ASNR 2021 Poster Abstracts

Frequency and profile of cognitive deficits in hospitalized COVID-19 patients undergoing inpatient rehabilitation

Abhishek Jaywant ¹ , Michael Vanderlind ¹ , George Alexopoulos ¹ , Chaya Fridman ¹ , Roy Perlis ² , Faith Gunning ¹

¹Weill Cornell Medicine. ²Massachusetts General Hospital/Harvard Medical School

Objective: Early reports and case series suggest cognitive deficits occur in some patients with COVID-19. We evaluated the frequency, severity, and profile of cognitive dysfunction in patients recovering from prolonged COVID-19 hospitalization who required acute inpatient rehabilitation prior to discharge.

Methods: We analyzed cross-sectional scores from the Brief Memory and Executive Test (BMET), administered in-person by a rehabilitation neuropsychologist, in a cohort of N=57 COVID-19 patients undergoing inpatient rehabilitation. The BMET comprises subtests that assess divided attention, motor speed, information processing speed, shifting/cognitive flexibility, immediate/working memory, delayed memory, recognition memory, and orientation. We calculated the frequency of impairment based on neuropsychologist diagnosis and by age-normed BMET subtests.

Results: In our sample, 43 patients (75%) were male, 35 (61%) were non-white, and mean age was 64.5 (SD=13.9) years. 48 (84%) were previously living at home independently. Only two patients had documented preexisting cognitive dysfunction; none had known dementia. Patients were evaluated at a mean of 43.2 (SD=19.2) days after initial admission. 50 patients (88%) had documented hypoxemic respiratory failure and 44 (77%) required intubation. 46 patients (81%) patients were found to have cognitive impairment, ranging from mild to severe. Deficits were common in immediate/working memory (55% of patients), shifting/cognitive flexibility (47% of patients), divided attention (46% of patients), and processing speed (40% of patients). Delayed memory and recognition memory were less frequently impaired (22% of patients each). Executive dysfunction was not significantly associated with intubation length or the time from extubation to assessment, prior documented delirium, psychiatric diagnosis, or pre-existing cardiovascular/metabolic disease.

Conclusions: Attention and executive functions are frequently impaired in medically-stable COVID-19 patients who require acute rehabilitation prior to discharge. Though interpretation is limited by lack of a comparator group, these results provide an early benchmark for identifying and characterizing cognitive difficulties after COVID-19. Cognitive dysfunction was not related to intubation length, and may not simply reflect a resolution of acute alterations in mental status from intubation/extubation and sedation. Further research is needed to determine the mechanism of these deficits, which may implicate regions that underlie attention and executive control including the frontal, parietal, and cingulate cortex as well as subcortical structures. Given the frequency and pattern of impairment, easy-to-disseminate interventions that target attention and executive dysfunctions may be beneficial to this population.

Increased functional impairment predicts progression to Mild Cognitive Impairment in cognitively unimpaired older adults: Setting the stage for the use of novel motor tasks in identifying preclinical Alzheimer’s disease

Andrew Hooyman ¹ , Tyler Rose ¹ , Michael Malek-Ahmadi ² , Sydney Schaefer ¹

¹Arizona State University, Tempe, USA. ²Banner Alzheimer's Institute, Pheonix, USA

Current strategies for treating and managing dementia-related disorders are focused on identifying disease risk prior to clinical diagnosis or symptom onset. While many biomarkers (e.g., brain amyloid, tau), genetic factors (e.g., Apolipoprotein E alleles), and cognitive factors (e.g., poor memory) are correlated with dementia progression, they do not show strong prediction of disease progression in cognitively intact older adults. However, recent empirical findings have suggested that functional motor behavior may be able to differentiate between progressors (those who progress to Mild Cognitive Impairment, MCI) and non-progressors (those who remain cognitively unimpaired [CU]) over time. To further test this premise, we utilized the National Alzheimer’s Coordination Center (NACC) database to retrospectively examine how baseline scores on the Functional Assessment Survey (FAS), a self-reported measure of independence on common activities of daily living, predicted eventual progression to MCI as determined by the Alzheimer’s Disease Neuroimaging Initiative (ADNI) criteria. The NACC database, like many other longitudinal observational studies in older adults, does not assess functional motor behavior; thus, we used the FAS as a proxy. To ensure comparable groups of progressors and non-progressors in our analysis, we generated a subset of NACC data that included equal numbers of progressors and non-progressors who were CU at baseline, had no pre-existing neurological conditions, and were matched on age, sex, education and baseline MoCA score. In the final sample of 322 subjects, progressors and non-progressors were similar with respect to age (72.14 vs 70.81 years), percent female (59% vs 57%), years of education (17 vs 17.83 years) and MoCA score (25.01 vs 24.95). On average, progressors were diagnosed with MCI 2.5 years post-baseline. A generalized linear model was used to predict conversion status with age, sex, education, total MoCA, and FAS as independent variables, and demonstrated that only FAS score predicted progression to MCI (OR = 1.35 [95% CI: 1.07 – 1.78], p=0.02). Results from this analysis indicates that greater functional impairment is a significant predictor of progression to MCI. Future research should further consider the application of motor-based assessments to predict dementia progression.

Emotional prosody recognition after right-hemisphere stroke – an fMRI study of functional brain reorganization

Anna Seydell-Greenwald ¹ , Katherine O'Connell ² , Abigail Marsh ² , Alexander Dromerick ³

¹Georgetown University Medical Center, Washington, DC, USA. ²Georgetown University, Washington, DC, USA. ³MedStar National Rehabilitation Hospital, Washington, DC, USA

Background: While language is a function generally attributed to the brain’s left hemisphere, a socially crucial aspect of language comprehension – emotional prosody recognition – is thought to be lateralized to the brain’s right hemisphere. Indeed, aphasias are more commonly observed after left-hemisphere lesions, whereas aprosodias are more frequent and severe after right-hemisphere lesions (Heilman et al., 1975; Borod et al., 1992; Witteman et al., 2011). Congruently, functional neuroimaging studies reveal left-lateralized frontotemporal activations during language tasks like sentence comprehension, but right-lateralized frontotemporal activations during prosody recognition tasks (Seydell-Greenwald et al., 2020).

Interestingly, patients with left-hemisphere lesions and aphasia often show increased right-hemisphere activation during language tasks (Turkeltaub et al., 2011), prompting aphasia researchers to investigate whether modulating this mirror-activation by means of non-invasive brain stimulation can augment aphasia therapy (Biou et al., 2019).

Objective: We asked whether an analogous mirror-activation occurs during emotional prosody recognition in the left hemisphere of patients with right-hemisphere lesions. If so, this might open up a new avenue for treating acquired emotion recognition impairments, which are linked to poorer social wellbeing after stroke (O’Connell et al., under review).

Methods: Ten survivors of right-hemisphere stroke affecting the MCA territory and 17 matched controls completed an fMRI task in which they alternatingly had to judge the emotional prosody or the semantic content of spoken sentences while being scanned at 3 Tesla.

Results: As expected, contrasting the BOLD signal in the semantic decision condition with that in a control condition (reverse speech) revealed left-lateralized frontotemporal language activations in both groups, and contrasting the emotion recognition with the semantic decision task revealed right-lateralized frontotemporal activations in the control group. Interestingly, such right-lateralized group activation was also evident for prosody recognition in the right-hemisphere stroke survivors. Quantification of lateralization at the individual level showed weaker right-lateralization and even left-lateralization in some stroke survivors. However, this change in lateralization was driven by a reduction of right-hemisphere activation rather than an increase in left-hemisphere activation. A direct group comparison revealed stronger right-hemisphere activations in the control group, but not stronger left-hemisphere activations in the stroke survivors.

Conclusion: In contrast to the commonly observed increases in right-hemisphere activations for language tasks in stroke survivors with left-hemisphere lesions, we found no evidence for increases in left-hemisphere activations during an emotional prosody recognition task in stroke survivors with right-hemisphere lesions. While a larger sample will be needed to confidently conclude that such increases are truly absent, their absence in this study suggests that they are not particularly common or strong. Thus, the results of aphasia treatment studies seeking to modulate contralesional brain activation may not easily translate into similar treatments for aprosodia following right-hemisphere stroke.

How much treatment does a person with aphasia need to get the best recovery? A novel method for investigating dose-response relationships in aphasia treatment

Sam Harvey ^1,2 , Miranda Rose ^1,2 , Michael Walsh Dickey ^3,2 , Marcella Carragher ^1,2

¹Latrobe University, Melbourne, Australia. ²CRE Aphasia Recovery & Rehabilitation, Melbourne, Australia. ³University of Pittsburgh, Pittsburgh, USA

Introduction: Aphasia is a common and debilitating long-term consequence of neurological injury (El Hachioui et al., 2013; Mitchell et al., 2020). Efficacious treatments that aim to enhance language skills and communication exist (Brady et al., 2016). However, little is known about how the amount of treatment (i.e., the dose) affects aphasia recovery (Harvey et al., 2020a) and there are few rigorous experimental studies that have directly compared the effect of different doses of treatment on aphasia recovery (Harvey et al., 2020b). Subsequently, there is little empirically-based clinical guidance for dose prescription in aphasia rehabilitation (Doogan et al., 2018).

Dose in behavioural neurological rehabilitation is a multidimensional construct (Hayward et al., 2020) comprising treatment schedule parameters (e.g., the duration and distribution of treatment sessions) and the activities undertaken within treatment sessions. In line with this multidimensional framework, we have developed a novel methodology and study protocol for an experimental clinical trial investigating dose-response relationships.

Aims: This Phase I study tests a novel dose prescription method, investigates the effect of personalising treatment dose, and explores dose-response relationships.

Methods: An online single-case experimental design using multiple baselines across doses with replication across six participants with chronic aphasia is being used to evaluate the effects of cued picture naming treatment (Kendall et al., 2014) on the acquisition and maintenance of picture naming skills.

A novel method for determining a theoretical maximum tolerable dose is being trialled to personalise dose relative to individuals’ baseline picture naming skills (speed and accuracy). All participants receive 15 45-minute sessions of treatment in which the number of pictures used and the number of naming attempts made throughout treatment (i.e., the dose conditions) are experimentally manipulated.

Treatment effects will be compared across different dose conditions to determine optimal dosing of cued picture naming treatment. Additionally, we will explore the impact of memory and attention, and self-reported measures of fatigue, motivation, and task difficulty on dose-response relationships.

Outcomes: It is hypothesised that 1) this novel dose prescription method will have immediate clinical utility; 2) treatment outcomes will be positively correlated with dose condition whereby higher doses are associated with greater acquisition and maintenance of naming skills; 3) dose-response relationships will be influenced by factors such as baseline picture naming skills and cognitive functions; and 4) dose tolerance will be mediated by fatigue, motivation, and task difficulty.

We will present preliminary data from the first participants to complete the study protocol.

Findings from this study may influence clinical practice by providing clinicians with a method for prescribing the appropriate dose of a common aphasia treatment for any person with aphasia suitable for this treatment. Researchers may be able to use this novel method to personalise the dose of other aphasia treatments.

Relationship between integrity of white matter tracts and severity of aphasia after stroke

Hyeonji Park, Sumin Jeong, Woo Been, Yumi Hwang, Yoonhye Na, Minjae Cho, Eunyeong Lee, Sung-Bom Pyun

Korea University College of Medicine, Seoul, Korea, Republic of Korea

Diffusion Tensor Imaging (DTI) technique is commonly used to map white matter tracts in stroke patients. Fraction anisotropy (FA) values which are defined by the degree of similarity among overall directions of molecules can be obtained with DTI and decrease in FA values of certain white matter tracts is believed to represent the degree of integrity of those tracts. We investigated the relationship between the integrity of white matter tracts and severity of aphasia after stroke. Among 160 stroke patients who had admitted department of Physical Medicine and Rehabilitation of Korea University Anam Hospital from January 2014 to November 2019, 71 stroke patients were finally included for analysis: 1) first ever stroke, 2) left hemisphere stroke, 3) DTI obtained within 90 days after the onset, 4) K-WAB performed within 4 weeks after the onset, 5) right-handed, and age above 20 and below 85 years. Among the white matter tracts of brain, we selected four tracts known to have a role in cognitive and linguistic functions were selected: arcuate fasciculus (AF), inferior longitudinal fasciculus (ILF), frontal aslant tract (FAT), and cingulum. The severity of aphasia was assessed by Wester Aphasia Battery (WAB) and FA values of each tract of bilateral hemisphere were measured by DTI tractography using DSI studio software. In our results, among the FA values of both sides of four WM tracts, AQ had a significant positive correlation with FA of left AF (r=0.518, p < 0.001), left FAT (r=0.324, p=0.006), and left ILF (r=0.416, p <0.001). Also, fluency score had positive correlation with FA of left AF (r=0.471, p < 0.001), left FAT (r=0.351, p=0.003), and left ILF (r=0.335, p=0.004); Comprehension had positive correlation with FA of left AF (r=0.432, p < 0.001) and left ILF (r=0.399, p=0.001); Repetition had positive correlation with FA of left AF (r=534, p=000), left FAT (r=0.246, p=0.039) and left ILF (r=450, p=000); Naming had positive correlations with FA of left AF (r=451m, p=000), left FAT (r=0.307, p < 0.001) and left ILF (r=374, p=001). In multiple linear regression analysis, FA value of left FAT was significant variable for fluency and comprehension score, and FA value of left AF, left FAT and Rt cingulum score was significant variables for repetition and naming, respectively. In conclusion, left AF was the most important tract affecting the overall severity of aphasia, especially the repetitive function. left FAT was considered to have an important role in fluency and speech production.

Effects of social comparative feedback on motor sequence learning and perceived competence

Allison Lewis, Rachel Bohnenkamp, Lynn Johnson, Dirk den Ouden, Sara Wilcox, Stacy Fritz, Jill Stewart

University of South Carolina, Columbia, USA

Positive social comparative feedback during motor skill practice leads to increased confidence and better retention of the motor skill.^1-3 However, prior research focused on the learning of upper extremity motor skills with gross movement requirements (i.e. general accuracy) rather than sequential movements with precise spatial and temporal demands, like what is required for many motor skills necessary for everyday life. The purpose of this study was to examine the effects of positive social comparative feedback on the learning of and confidence in the performance of a motor task with spatial and temporal demands (a joystick-based motor sequence task). Forty-eight non-disabled adults (mean age 25.4; 33 females) completed practice of the motor task for two sessions, 24 hours apart. Participants were randomized to one of three different feedback groups: a control group (CO; received feedback that they completed the block of practice), a response time only group (RT only; received feedback on their response time to complete a block of practice) or a positive feedback group (RT+POS; received feedback that their response time was better than the average of others, i.e. positive social comparative feedback). The group-specific feedback was provided after each block of practice on Day 1. During practice (Day 1) and retention (Day 2), participants practiced two types of 8-target sequences (random sequence, repeated sequence) with a joystick in alternating order for a total of 140 sequences on Day 1 and 60 sequences on Day 2. The primary performance outcome was the response time to complete an 8-target sequence and self-reported perceived competence. For practice and retention, we found a main effect of group, block, and sequence type (p≤0.002). Post hoc testing revealed that all groups showed faster response times over practice (Day 1) that were maintained at retention (Day 2) and that response time to complete a sequence was faster for repeated than for random sequences on both days. The CO group and RT+POS group showed faster response times over practice and at retention than the RT only group, despite no differences between groups on baseline performance. In addition, there was a main effect of feedback group on competence (p=0.033), where the CO and RT+POS groups reported greater perceived competence over practice and at retention than the RT only group. Providing response time feedback without context was less beneficial for skill acquisition, learning and competence than no performance feedback or performance feedback with social comparative context. Our results suggest that providing performance feedback without context during motor practice for rehabilitative purposes may be detrimental to skill learning and competence.

Improvement and Retention of Dual-Task Reactive Stepping in People with Parkinson's Disease and Age-Matched Controls

Andrew Monaghan ¹ , James Finley ² , Shyamal Mehta ³ , Daniel Peterson ^1,4

¹Arizona State University, Phoenix, USA. ²University of Southern California, Los Angeles, USA. ³Mayo Clinic, Scottsdale, USA. ⁴VA Health Care Center, Phoenix, USA

The ability to elicit quick and effective movements after a sudden loss of balance, such as a reactive step, is critical in guarding against a fall. People with Parkinson's Disease (PD) demonstrate a compromised ability to elicit robust reactive steps compared to age-matched controls, but this ability improves with repeated practice. Despite the improvements shown with reactive step practice, it is unclear if reactive stepping can be enhanced when performing a concurrent cognitive task. This is important because falls often occur when attention is divided, and such circumstances of walking with divided attention are ubiquitous in daily life. Therefore, this study aims to determine whether people with PD and healthy controls can improve reactive stepping while performing a concurrent cognitive task and whether improvements are retained 24-hours later. Sixteen people with PD (Hoehn & Yahr 2-3) and thirteen age-matched healthy controls (HC) experienced support surface translations to induce backward reactive stepping while simultaneously performing an auditory Stroop task. Participants returned the following day to reassess dual-task reactive stepping performance. Outcome measures included cognitive reaction time to stimuli from the Stroop task, the onset of the tibialis anterior (TA) activation, and step length. Secondary measures included step onset latency and the margin of stability (MOS) at foot contact. Both the PD and healthy control (HC) groups increased step lengths after practice on day one (p <0.001). A significant group by time interaction was observed for cognitive reaction time, indicating that the HC group exhibited more substantial reductions in reaction time than the PD group (p <0.001). No improvements were observed for step latency, the MOS, or TA onset after practice on day 1. Retention was assessed for step length and cognitive reaction time as these were the only measures to demonstrate significant improvement on day 1. A subtle but significant retention effect was observed for step length across all participants (p = 0.05), indicating that step lengths were on average 13 mm larger at the start of day two compared to day one. Cognitive reaction times were also 83 ms quicker for all participants on day two (p <0.001). The degree of retention was not significantly different across groups as no group by task interactions were observed. This study provides preliminary evidence for the effectiveness of dual-task reactive step training in people with mild PD and healthy controls. Since participants learned to increase step length without compromising cognitive reaction times, this suggests that improvements in reactive stepping do not require individuals to shift attention toward the postural task. The retention of improvements in step length demonstrates the potential application of dual-task reactive step training in rehabilitation strategies.

STEGA: an iPad app to measure precision drawing and handwriting

Benjamin Philip ¹ , Fuhai Li ¹ , Elizabeth Hawkins-Chernof ¹ , Victoria Swamidass ² , Igor Zwir ¹

¹Washington University School of Medicine, St. Louis, USA. ²PlatformSTL, St. Louis, USA

Clinicians and researchers cannot easily assess fine motor control (hand and fingers) and handwriting because current methods are expensive, slow, and/or subjective. We developed and validated an iPad app (STEGA, Standardized Tracing Evaluation and Grapheme Assessment) for rapid, quantitative, scalable assessment of the fine motor control skills that underpin handwriting. We hypothesized that STEGA drawing would provide a partial but significant predictor of handwriting performance (r² = 0.25) in school-age children. 57 typically developing children ages 9-11 used the STEGA app to draw 30 symmetrical Abstract shapes. Participants also completed a benchmark assessment, the Evaluation Tool of Children’s Handwriting-Cursive (ETCH-C). STEGA was able to predict ETCH-C letter legibility with r² = 0.44 using support vector machine regression, which outperformed machine learning methods. STEGA app development included dynamic onscreen guides to minimize cognitive load, experimenter control over the order and width of drawing shapes, experimenter access to full data at 50 Hz and 0.5 mm spatial precision, and immediate onscreen summary statistics. STEGA was well-accepted by children: all participants completed the app in 4-11 minutes (mean 6.7, SD 1.3), and 93% provided neutral-to-positive feedback. Therefore, STEGA provides a rapid objective assessment of fine motor control with relevance to handwriting. Future studies will need to validate STEGA in larger samples across multiple age groups, to establish normative data and allow machine learning to capitalize on the data’s structure. STEGA’s high-precision low-cost data will lower the barrier of entry to research in fine motor control and its rehabilitation. Moreover, objective precision data could potentially improve rehabilitation by allowing earlier detection of developmental disorders in children (e.g. dyspraxia), and allow at-home tracking of progressive disorders such as Parkinson’s Disease.

Reach Peak Velocity Affected by Task Complexity and Variability in Grasp Strategy During Prosthesis Use

Bennett Alterman ¹ , Saif Ali ¹ , Emily Keeton ¹ , William Hendrix ² , Jade Lee ³ , John Johnson ¹ , Katrina Binkley ¹ , Lewis Wheaton ¹

¹Georgia Institute of Technology, Atlanta, USA. ²Kenney Orthopedics, Lexington, USA. ³Hanger Clinic, Dallas, USA

Examining differences in behavioral and functional performance provides vital information about sensorimotor reorganization at amputations of different levels. However, kinematic performance may depend on several factors including task complexity and grasp strategy variability by the prosthesis user. This study examines differences in behavioral and functional outcome measures in participants completing goal-directed reach and grasp motor actions, and provides a preliminary basis for evaluating grasp strategy variability as a mechanism of differentiation in motor learning. Here, intact participants completed a kinematically simple and complex reach and grasp task using either a transradial (n=17) or partial-hand (n=16) prosthesis simulator. The simple task involved the lateral translation of a metal disk between two points. The complex task involved the rotation and translation of a marker from a horizontal position to a vertical position. We hypothesized that the complex task would lead to greater grasp strategy variability, and that partial-hand prosthesis users would exhibit higher reach peak velocities than transradial prosthesis users. We find that in the complex task, partial-hand users stratify into uniform graspers and variable graspers (used the same grasp strategy for <85% of grasps). Kinematic results show a main effect of both group and trial bin, where there were significant differences between the transradial and partial-hand groups in later trial bins. Furthermore, partial-hand users show increases in reach peak velocity over time, while transradial users do not. In the simple task, there is no stratification in grasp strategy. Kinematically, there is a main effect of trial bin driven by partial-hand users. This work forms the foundation for examination of how level of amputation may affect grasp strategy and kinematic outcomes during rehabilitation.

Hebbian-type stimulation of premotor cortex. Development of a new treatment approach for impaired hand function after stroke

Paul Christian ^1,2 , Scott Shaeffer ¹ , Cathrin Buetefisch ¹

¹Emory University, Atlanta, USA. ²Dresden University, Dresden, Germany

There is a need for more effective rehabilitation strategies to improve hand function post-stroke. Because aspects of stroke motor recovery and motor learning share similarities of their underlying mechanisms, a better understanding of means to enhance them will inform the development of more effective treatment protocols. Repetitive transcranial magnetic stimulation (rTMS) when applied to primary motor cortex (M1) in a specific temporal relationship to hand training (Hebbian-type rTMS) enhances motor learning in healthy adults (Buetefisch et al., 2014) and stroke patients (Revill et al., 2020) through its effect on M1 plasticity. However, M1 may not be a salient target when M1 function is disrupted after stroke. Premotor cortex (PMC) may serve as a target for rTMS to promote recovery of hand function because reorganization of PMC supports motor recovery and becomes critical with more extensive injury to M1. Here we test the hypothesis that Hebbian-type rTMS applied to PMC during motor training will enhance motor learning. Outcome measures are improvement in hand motor kinematics (peak velocity and reaction time of ballistic wrist extension movements) and motor learning-related reorganization in M1 (reflected in increased M1 excitability). To test our hypothesis, five different rTMS protocols were tested in 10 healthy right-handed adults (6F, age= 56.0±5.73 years) using a cross-over study design. On five days participants performed an auditory paced hand motor training while receiving Hebbian-type rTMS to either ipsilateral or contralateral PMC or control rTMS protocols. Participants performed ballistic wrist extension movements to move a cursor to a target displayed on a screen. The cursor represented real-time feedback about the position of the hand which was measured by a two–dimensional gyrometer mounted on the dorsum of the hand. Target location changed with respect to the distance from the home position which reflected the position of the hand at rest. Subthreshold, 0.1 Hz rTMS was applied to PMC at 50 ms prior to the movement onset. As a control, rTMS was also applied to sensory cortex and to PMC at +400 ms after movement onset, or at minimal intensity (sham). Hand kinematics and M1 excitability were measured before, immediately after training and again 60 minutes after training. For M1 excitability, stimulus response curves (SRC) and short-interval cortical inhibition (SICI) were measured in the hotspot of a muscle supporting the training movements. Preliminary results demonstrate different effects of the rTMS intervention on hand kinematics depending on the stimulation site, intensity and time of stimulus. Results for the effects on SRC and SICI are pending. Extending findings of previous studies on the effect of rTMS on motor kinematics, we demonstrate now that Hebbian-type stimulation of PMC during motor training is feasible and produces specific effects with respect to location, timing and intensity.

A rapid Upper-Extremity Fugl-Meyer assessment tool for patient selection in rehabilitation trials

Abigail McGeorge, Leo Cekus, Winston Byblow, Cathy Stinear

University of Auckland, Auckland, New Zealand

Background: The upper-extremity Fugl-Meyer assessment (UE-FMA) is commonly used to assess motor performance after stroke. Clinical trials of UE rehabilitation often select patients based on their total UE-FMA score, typically excluding patients with high scores to avoid ceiling effects.¹ Patient selection might be more efficient if assessments could be completed rapidly and remotely. The aim of this study was to identify a subset of UE-FMA items that could accurately identify patients above and below a cut-off for inclusion in a trial enrolling patients within 7 days post-stroke.

Methods: Retrospective data were analysed from 447 patients (189 (42%) female, mean age 70 (range 18 – 99 years), 393 (88%) ischaemic stroke). Baseline UE-FMA scores were obtained mean 5 days (range 1 – 14 days) post-stroke, and binarised at several cut-off points systematically varied around the desired score for inclusion. A Classification and Regression Tree (CART) analysis was used to identify UE-FMA tasks that correctly binarised patients’ total UE-FMA score. Child node size and maximum tree depth were varied, Gini was used to maximise homogeneity of child nodes, trees were pruned, and 10-fold cross-validation was conducted. CART analysis was performed with all UE-FMA tasks available, and repeated after removal of tasks that have poor reliability or construct validity² or are not feasible for remote assessment by video call. The target cut-off score was 51, to exclude patients with a baseline score of 50 or more without the need for a full UE-FMA assessment. The decision trees produced by CART analysis could make two types of errors: incorrectly predicting a score greater than 51 (‘wrongly excluded’) and incorrectly predicting a score ≤ 51, thus requiring an unnecessary full UE-FMA.

Results: CART analysis of data binarised at 51 (0-51, 52-66) produced a tree that uses three UE-FMA tasks to categorise a patient as having a total UE-FMA score ≤51, or >51. These were shoulder flexion to 90°, forearm pronation-supination with shoulder at 0° and elbow at 90°, and holding a pen between index finger and thumb. Overall accuracy with cross-validation was 94.4% (SE 1.2%). Using the tree would have wrongly excluded five patients (1.1%) and unnecessarily required a full UE-FMA for 22 patients (4.9%). The same tree was found to be 94.9% accurate at 12-weeks post-stroke (n=353) and 93.3% accurate at 26-weeks (n=176).

Discussion: This proof-of-concept shows it may be possible to accurately select patients for UE rehabilitation trials using a rapid assessment tool either in person or remotely. Doing so may be useful in the context of the COVID-19 pandemic, and for screening potential participants before they travel to a research facility for full assessment. Further work is needed to prospectively validate the decision tree, and to develop other trees for different UE-FMA scores.

Rapid and Remote Categorisation of Upper Limb Motor Outcome after Stroke

Harry Jordan ¹ , Joia Che ² , Winston Byblow ¹ , Cathy Stinear ¹

¹University of Auckland, Auckland, New Zealand. ²Monash University, Melbourne, Australia

Background: The Action Research Arm Test (ARAT) is the only functional assessment of upper limb activity currently recommended by the Stroke Recovery and Rehabilitation Roundtable. The ARAT score has been used to classify upper limb functional outcome after stroke in one of three,¹ four,² and five³ categories. The COVID-19 pandemic has encouraged the development of assessments that can be performed quickly and remotely. To our knowledge there are no reported attempts to use a subset of ARAT tasks to rapidly categorise current functional outcome either in-person or remotely. The aim of this study was to derive and internally validate decision trees for categorising upper limb motor outcome at the late sub-acute and chronic stages after stroke using a subset of ARAT tasks. A rapid remote ARAT assessment could provide researchers with a method for categorising upper limb function when circumstances prevent in-person assessment.

Methods: This study retrospectively analysed 3-month post-stroke ARAT scores obtained in-person from 333 patients who were recruited within the first 3 weeks post-stroke [180 males (54%), median age 70 (range 18–98) years]. In-person ARAT scores were used to categorise patients’ 3-month upper limb outcome using each of the three previously published classification systems. Scores from ARAT tasks were then used in classification and regression tree (CART) analyses to determine whether a subset of tasks could accurately categorise upper limb outcome. Only tasks that were feasible for remote assessment were included in analyses. Minimum child node size was 10, maximum tree depth was four, Gini was used to maximise homogeneity of child nodes, and 10-fold cross-validation was conducted.

Results: CART produced decision trees based on 2 to 4 ARAT tasks, all from the grasp and pinch subscales. The overall accuracy of the cross-validated decision trees ranged from 87.7% (SE 1.0%) to 96.7% (SE 2.0%). Accuracy was highest when classifying patients into one of three outcome categories, and lowest when using five categories. The decision trees remained accurate for patients assessed at 6 months post-stroke (n = 158, accuracy 83.5% - 91.8%).

Discussion: This proof-of-concept study indicates that 2 – 4 ARAT tasks can accurately categorise functional upper limb outcomes after stroke. Rapid assessments that can be completed remotely via videocall may be useful in the context of the COVID-19 pandemic. We are currently investigating the accuracy and feasibility of categorising current upper limb function using rapid remote ARAT assessments compared to full in-person ARAT assessments with late subacute and chronic stroke patients.

Instrumented measurement of contracture and spasticity in the lower limbs of people with CP undergoing activity-based training: a systematic review

David Yang ¹ , Caitlin Hurd ² , Jaynie Yang ² , Diane Lorenzetti ¹ , Elizabeth Condliffe ¹

¹University of Calgary, Calgary, Canada. ²University of Alberta, Edmonton, Canada

Background: Activity-based training programs are increasingly implemented as part of the rehabilitation strategies for people with cerebral palsy (CP), particularly to address activity limitations and barriers to participation. Contracture and spasticity are body structure and functional impairments that the majority of individuals with CP experience, which may underlie activity limitations. Instrumented outcome measures can often reveal the relative importance of contracture versus spasticity whereas this is can be challenging or impossible with clinical measures. We present a systematic review of the impact of activity-based interventions on contracture and/or spasticity evaluated with instrumented measures in the lower extremities of people with CP.

Methods: Relevant literature was obtained from 7 databases: Embase, PubMed, Scopus, CINAHL, SPORTDiscus, Cochrane CENTRAL Register of Controlled Trials, and PEDro. Two independent reviewers determined study eligibility. Strength of evidence and methodological quality were assessed using the Oxford Centre of Evidence-Based Medicine Levels of Evidence and Physiotherapy Evidence Database (PEDro) scale.

Results: Of 3805 articles identified, 20 met all inclusion criteria. Ten articles each were rated as level 2 or 3 evidence on the OCEBM scale. A wide range of study quality ranging from PEDro score 3 to 8 was observed. Interventions included: strength training with or without dedicated stretching, gait training, sports, general exercise. Contracture or spasticity were not the primary outcome in several studies.

Instrumented measures of contracture included passive range of motion and passive joint stiffness measured with instrumented goniometers or dynamometers and the muscle compressibility ratio measured with ultrasonography. Instrumented measures of spasticity included reflex stiffness measured with dynamometers, as well as H/M ratio measured with nerve stimulation in conjunction with electromyography (EMG). Several instrumented measures using instrumented goniometers and dynamometers did not clearly differentiate contracture and spasticity.

Nine of the 16 studies that measured contracture outcomes noted improvements. Of the 4 studies that measured spasticity outcomes, all 3 that evaluated reflex stiffness saw no changes while the study that reported H/M ratio reported improvements. Only 1 of the 4 studies that used outcomes that did not differentiate contracture and spasticity saw improvements. Most notably, no studies observed increases in contracture or spasticity. Interestingly, all 4 studies examining combined stretching and strength training programs reported improvements in contracture. A meta-analysis was not pursued as few studies were of adequate quality and reported enough detail.

Conclusion: While activity-based training is often implemented to target other goals, it also has the potential to manage contracture and spasticity. In particular, combined stretching and strength training appears to improve contracture, while other interventions result in no negative impacts on contracture or spasticity. Furthermore, instrumented measures that distinguish between contracture and spasticity allow therapies to be tailored to the specific impairment.

Impaired visuomotor integration during rapid reaching and interception in children with cerebral palsy

Deborah A. Barany ¹ , Owais Ahmed Khan ¹ , Ana Gómez-Granados ¹ , Tarkeshwar Singh ² , Christopher M. Modlesky ¹

¹University of Georgia, Athens, USA. ²The Pennsylvania State University, University Park, USA

Accurate goal-directed movements require rapid integration of incoming sensory information with motor commands. Children with spastic cerebral palsy (CP) have difficulties performing visually-guided upper-limb motor tasks (e.g., catching a ball, grasping a cup), but the extent to which these functional deficits are related to impaired visuomotor integration, independent of pure visual or motor dysfunction, remains unclear. To explore this relationship, we examined rapid goal-directed reaching and interception movements in ambulatory children with mild spastic CP (n = 12, 7 F, 9.08 ± 2.13 y) and typically developing (TD) controls (n = 7, 4 F, 7.68 ± 1.98 y). Relative to reaching toward stationary targets, intercepting moving targets requires continuous incorporation of new visual motion signals to improve spatial and temporal precision throughout the movement. Thus, we hypothesized that specific visuomotor integration deficits in children with CP would result in impaired performance in the interception task relative to the reaching task.

We recorded limb kinematics as seated participants grasped a robotic manipulandum (KINARM End-Point Laboratory, KINARM, Kingston, Ontario, CA) with one hand to perform whole-arm, horizontal plane reaching movements in response to visual targets on a virtual display. On each trial, a target either (a) appeared on the right or left side of the visual workspace and remained in the same position (Reaching), or (b) appeared in the center and then moved leftward or rightward across the workspace at a constant Euclidean velocity (Interception). Participants completed four blocks of Reaching and Interception trials with the dominant and non-dominant arm (16 blocks total). The velocity of the moving target during Interception (and corresponding time-to-hit target during Reaching) varied for each block of trials.

Across Reaching and Interception blocks, children with CP had significantly lower success rates (CP: 46 ± 6%; TD: 64 ± 7%), slower tmovement times (CP: 282 ± 25 ms; TD: 225 ± 18 ms), and longer movement path lengths (CP: 103 ± 5 mm; TD: 92 ± 5 mm) than TD children. As expected, success rates decreased during blocks with less time-to-hit the target, for both Reaching and Interception. Interestingly, for non-dominant (more affected) arm movements, the decrease in success rate with decreasing time-to-hit during Interception (e.g., increasing movement velocity) was significantly larger in children with CP relative to TD children. In contrast to Reaching, in which reaction and movement times tended to scale with task difficulty for both CP and TD children, children with CP showed relatively similar reaction and movement times across Interception blocks. Together, these results suggest that the ability to integrate visual information about target velocity to plan and adjust ongoing movement is impaired in children with CP.

A new method for extracting neural correlates of movement in people with movement disorders: A study of ipsilateral control for bimanual function in pediatric hemiplegia

Disha Gupta ^1,2 , Alexandre Barachant ³ , Jason Carmel ^4,5 , Kathleen Friel ^3,6

¹National Center for Adaptive Neurotechnologies, Albany, USA. ²United States Department of Veterans Affairs, Albany, USA. ³Burke Neurological Institute, White Plains, USA. ⁴Columbia University Vagelos College of Physicians and Surgeons, New York City, USA. ⁵NewYork-Presbyterian Morgan Stanley Children’s Hospital, New York City, USA. ⁶Weill Cornell Medicine Medical College, New York City, USA

Objective: In early brain injury, such as in unilateral hemiplegic cerebral palsy (UHCP), the uninjured hemisphere has increased connectivity with the affected hand^1,2. This has been extensively probed via diffusion tensor imaging³ and by transcranial magnetic stimulation⁴ with the hand at rest. However, physiological connectivity during active movement is less explored. More importantly, it remains unclear how this ipsilateral activation relates to bimanual movement. The necessary neurophysiological assessment remains challenging in people with movement disorders, especially children. This is mainly due to the weakened neural responses that have atypical and heterogeneous spatio-temporal features², and the difficulty in identifying onset and offset of movement, even with EMG activity, due to unavoidable movement artifacts. We describe a novel analysis routine for objective and optimal feature extraction that overcomes these constraints and thereby helps in the investigation of ipsilateral control of bimanual function.

Method: We combine cortical (EEG) and muscular (EMG) signals acquired during a game-based movement execution paradigm in which EMG signals control the game parameters in real time. EEG and EMG are simultaneously cross-decomposed with advanced subspace-decomposition techniques for objective feature discrimination. We validate and compare the method with two conventional methods (SpoC⁵ and CSP⁶) on data from 6 neurotypical people (44 measurements from multiple sessions). Next, we apply it to data from 15 children (age: 7.02-16.10 years.months, 8f/7m) with congenital UHCP. All data were collected with informed consent; protocols were approved by the responsible IRBs.

Results: In the validation phase, our method was consistent with conventional methods in extracting hand-movement-related suppression of EEG sensory-motor rhythms⁷: with SpoC (χ²= 0.02, p= 0.90); with CSP (χ²= 0.96, p= 0.33). It was significantly better in identifying features objectively: versus SPoC (χ²= 23.72, p= 1.11e-06); versus CSP (χ²= 19.06, p= 1.26e-05). This is a key requirement for UHCP analysis. In UHCP data, we first extracted the spatial patterns for the unaffected-hand. They were typical contralateral responses in the uninjured motor cortex, similar to the validation phase. Next, we extracted the pattern for the affected hand, and determined whether it was predominantly contralateral or ipsilateral; 12 of 15 participants had a dominant ipsilateral response during affected-hand movement. The relative strength of the affected- and unaffected-hand movement at the uninjured cortex was significantly related to the bimanual hand function scores as measured by Assisting Hand Assessment Score⁸ (Spearman's ρ= 0.67, p= 0.0089, ρ²= 0.44).

Conclusion: among subjects with early brain injury who exhibit ipsilateral control, more evenly balanced representation of affected and unaffected hand in uninjured cortex is indicative of better bimanual hand function. Similar assessment of a larger group could help determine whether encouraging ipsilateral control of the affected hand could improve bimanual function.

A comparison of mean truncal accelerometry during upper extremity rehabilitation exercises in individuals with post-stroke hemiparesis and healthy controls

Edwin Dang ¹ , Noah Balestra ¹ , Ania Busza ²

¹University of Rochester, Rochester, USA. ²University of Rochester Medical Center, Department of Neurology, Rochester, USA

Background: Studies on upper extremity (UE) movements in individuals who have weakness from stroke have found altered movement patterns as compared to healthy controls. For example, compensatory movements of the trunk and upper arm during arm extension has been reported in patients with post-stroke hemiparesis. Furthermore, it has been suggested that such compensatory movements during post-stroke rehabilitative exercise sessions may impair motor recovery. Developing methods for automatically detecting such compensatory movements would be helpful for further expanding our understanding of the frequency, significance, and effect of such movements.

Objective: To use wearable sensors to assess the relative amount of truncal movement, as measured by mean acceleration of healthy and stroke subjects during UE exercises.

Design: As part of a larger study aiming to use wearable sensors and machine learning strategies to classify and quantify UE exercises, we have been collecting movement data from healthy controls and subjects with UE weakness due to recent (<4 weeks) unilateral stroke. Subjects are asked to perform a variety of functional upper extremity activities while seated, including cup stacking, wiping, radial motion, and rolling a rolling pin while wearing 7 sensors that collect accelerometry and gyroscopy data. Three sensors are placed on each arm (one on the dorsum of the hand, one on medial forearm, one overlaying three biceps of each arm, and one sensor on the sternum). During the exercise sessions, the subjects are video recorded to obtain information about the time and duration of activities performed. In this analysis, z-axis accelerometry data from the sensor placed on the chest was synchronized to the video time stamps and then accelerometry data from the z-axis (orthogonal to the chest wall) was used to assess truncal movement. The mean absolute accelerometry during the time the subject was performing the rolling pin exercise was calculated for all subjects. The mean across all healthy participants was then compared with the mean of all stroke subjects.

Results: Preliminary results with our first 9 patients (4 healthy, 5 individuals with recent stroke) suggest a difference in mean accelerometry between healthy subjects (0.299254 g) and stroke (0.431259 g). Healthy subject data ranges from 0.174382 g to 0.295307 g, whereas stroke subject data ranges from 0.080184 g to 0.717733 g.

Conclusion/Future Direction: Our preliminary results suggest that mean accelerometry of the trunk during a bilateral arm exercise is larger in patients with hemiparesis due to recent stroke than in healthy controls. Further analyses will include additional subjects and investigate whether this trend is seen in other arm exercises, and whether the effect is increased at specific phases of each exercise.

Using accelerometry to quantify motor traits of children with and without autism

Jeff Konrad, Natasha Marrus, Catherine Lang

Washington University, St. Louis, USA

Objectives: Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by deficits in social communication and restricted, repetitive behavior and is associated with multiple neurodevelopmental liabilities¹. Each liability can be thought of as a trait that is continuously distributed in the population. Impaired motor traits are highly common in ASD, strongly correlated with other ASD traits, and are an early-emerging sign of risk for ASD²; thus, measurement of motor traits with wearable technology may be useful for evaluating profiles of adaptive function and enhancing early screening. Our objective was to determine the feasibility of bilateral wrist accelerometry for prolonged periods in children with and without ASD, and to compare the distributions of motor and non-motor traits using accelerometry sensors and parent-reported questionnaires.

Methods: We recruited 48 children with and without ASD between the ages of 3 and 10. The children wore 3-axis accelerometers on each wrist. Parents were instructed to have the child wear the sensors for 12 waking hours on 2 separate days. Parents completed questionnaires measuring coordination, hyperactivity, and autistic social impairment. In addition to the 12-hour period, the maximum, median, and minimum hours of waking activity were extracted from the time series data in order to compare children across a range of real-world activity levels. Sensor variables that quantify motor behavior were calculated for each of these time segments. Analyses explored descriptive statistics and relationships between sensor variables, questionnaire scores, recorded day, age, and gender.

Results: The average wear time per day was 11.88 hours. The vast majority of participants tolerated the sensors well and no adverse events occurred. The sensor variables did not significantly differ across the two recording days. This was true when school was in session (school days compared to weekends and when school was out of session (e.g. during COVID-19 lockdowns or summer break). The parent-reported indices of coordination, hyperactivity, and autistic social impairment and most accelerometry variables did not vary with age. Two exceptions were simultaneous movement hours and sample entropy which were both negatively correlated with age. Neither the parent-reported indices nor the sensor variables differed between genders.

Conclusion: Collecting movement data from both wrists of children ages 3 – 10 for 12 waking hours is feasible. Similar values across separate recorded days suggests that school, sports, or other daily activities may not influence the sensor variables in this age group. The moderate relationships found with age indicate that older children tended to have more regular, but less bilateral upper limb movements. Results and conclusions are preliminary and further analyses are ongoing.

Relating global cognition and upper-extremity motor skill retention in individuals with mild-to-moderate Parkinson disease

Jennapher Lingo VanGilder ¹ , Cielita Lopez-Lennon ² , Serene Paul ^2,3 , Leland Dibble ² , Kevin Duff ^4,5 , Sydney Schaefer ¹

¹Arizona State University, Tempe, USA. ²University of Utah, Salt Lake City, USA. ³The University of Sydney, Sydney, Australia. ⁴University of Utah Health Sciences, Salt Lake City, USA. ⁵University of Utah Hospital, Salt Lake City, USA

Background: Cognitive impairment has been linked to poor motor learning and rehabilitation outcomes in older adult and stroke populations, but whether cognition interferes with motor skill learning in individuals with Parkinson disease (PD) remains unexplored. The purpose of this secondary data analysis from a recent clinical trial (NCT02600858) was to determine if global cognition was related to nine-day skill retention after upper-extremity motor training in individuals with PD.

Methods: Twenty-three participants (mean±SD age=71.07±6.88) with a confirmed diagnosis of idiopathic PD completed three consecutive days of training on an upper-extremity task. For the purposes of the original clinical trial, participants trained either on or off their dopamine replacement medication. Baseline, training, and shorter-term (48-hour) retention data have been previously published. Global cognitive status was evaluated using the Montreal Cognitive Assessment (MoCA). Participant age, baseline performance, MoCA score, and group (medication on/off) were included in a general linear regression model to predict longer-term (nine day) follow-up performance. Baseline and follow-up performance were assessed for all participants while on their medication.

Results: Linear model results indicated that MoCA score predicted nine-day follow-up performance (ß=-3.02; 95% CI [-5.06, -0.97], p=0.006), such that individuals with higher scores were faster than those with lower scores. Participant age (ß=-0.21; 95% CI [-0.89, 0.47], p=0.53) and medication status group (ß=2.04; 95% CI [-2.24, 6.32], p=0.33) did not. Baseline motor task performance (ß=0.47; 95% CI [-0.03, 0.98], p=0.07) trended towards significance.

Discussion: Results of this secondary analysis indicate that global cognitive status (which was not originally considered nor stated as a dependent variable in the primary outcome report) was a significant predictor of motor task performance well after training had been completed (nine days later), regardless of whether training had occurred “on” or “off” dopamine replacement medication. These results align with previous work that suggest cognitive impairment may interfere with motor learning, and that cognitive testing could be a useful tool to provide prognostic insight into an individual’s responsiveness to motor rehabilitation for a number of clinical populations.

Alterations in upper limb muscle synergies underlying coupling of end-point force post stoke during isokinetic force generation

Jeong-Ho Park ¹ , Hangil Lee ¹ , Jinsook Roh ² , Hyung-Soon Park ¹

¹Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Korea, Republic of Korea. ²University of Houston, Houston, USA

To perform a given motor task, muscles in the human body coordinate with one another. Intermuscular coordination can be quantitatively described as weights of muscles that form muscle synergies (i.e., modules of the coordination) and activation of the synergies. Stroke, which induces motor impairments in the affected limbs, results in alteration of muscle synergies, but different mechanisms of alteration have been reported depending on the motor tasks performed by a subject. A comparison of the different patterns of muscle synergies post stroke across a variety of motor tasks not only improves our understanding of how stroke affects intermuscular coordination but also enables systematic development of rehabilitation training for the stroke-affected intermuscular coordination.

We evaluated muscle synergies underlying the shoulder and elbow joints of stroke-affected upper limbs during isokinetic force generation tasks, whose results have been unknown. Chronic stroke survivors and neurologically intact participants (as control subjects) intended to move their hands straight along six orthogonal directions (e.g. anterior, posterior, medial, lateral, superior, and inferior) at a constant speed and to generate end-point forces at maximum effort toward the movement directions. In addition to the muscle synergies, we evaluated the coupling of force components in different directions to quantify how accurate the control of end-point force direction was in both groups.

Compared to the control group, the stroke survivors showed coupling among the force components, which indicated that the end-point forces were less accurately controlled in the stroke-affected limbs. Specifically, the end-point forces along the anterior, medial, and inferior directions were coupled with one another while the posterior, lateral, and superior forces were coupled. In terms of muscle synergy, in the stroke group, we observed that activation of shoulder-related synergies was altered differently depending on whether the elbow joint was flexed or extended. The altered activation of muscle synergies tended to explain the coupling of the end-point force.

These results suggest that the coupling of the end-point force during isokinetic force generation can be utilized as an indicator of abnormal coordination between muscles underlying the shoulder and elbow joints. In addition, training independent activation of muscle synergies, which involve excitation of muscles around the elbow or shoulder joint, can improve motor control performance in the stroke-affected limbs such as reduction of the end-point force coupling.

Accuracy of the PREP2 algorithm in predicting upper limb functional capacity in a United States population with first ever stroke

Jessica Barth ¹ , Kimberly Waddell ¹ , Marghuretta Bland ^1,2,3 , Catherine Lang ^1,4,3

¹Washington University in St. Louis, Program in Physical Therapy, St. Louis, USA. ²Washington University in St. Louis, Program in Occupational Therapy, St. Louis, USA. ³Washington University in St. Louis, Neurology, St. Louis, USA. ⁴Washington University in St. Louis, Program in Occupational Therapy2, St. Louis, USA

Introduction: The ability to predict upper limb (UL) outcomes for individuals after stroke facilitates treatment selection, discharge planning, and goal setting. PREP2 is an algorithm, developed in New Zealand that allows for prediction of UL functional capacity at 3 months post stroke based on measures taken within the first week with 71% accuracy. The algorithm classifies individuals into one of four expected categories: Excellent, Good, Limited, or Poor. Despite its accuracy and ease of use, potential challenges of applying PREP2 to persons with stroke in the US include: different timing and structure of rehabilitation services provided, and lack of routine access to the neurophysiological measure. Thus, the objective of this study was to determine the accuracy of PREP2, using clinical measures only, on a sample of persons with first ever stroke in the US. It was hypothesized that algorithm accuracy would fall in a range of 70-80%.

Methods: Adults with first ever stroke (N=49) with motor deficits of the upper limb were recruited from a local hospital and followed longitudinally, as part of an ongoing observational cohort. Dependent variables captured between 24 hours out to two weeks post stroke and entered into the algorithm were: age, degree of UL paresis (SAFE score) and stroke severity (NIHSS total score). The dependent variable predicted was category of UL functional capacity, defined by ranges of the 3-month ARAT score. Predictions were evaluated at 12 weeks (84 ± 7 days) after stroke. Overall accuracy, specificity, sensitivity, positive predictive value (PPV), and negative predictive value (NPV) of the algorithm, were calculated using a 4x4 contingency table.

Results (Poster): Data from 49 individuals were analyzed (57% male, 88% ischemic stroke, age = 65 ± 8.56 years). The overall accuracy (61%) and weighted kappa (62%) were significant. Sensitivity was high for the Excellent (95%) and Poor (81%) PREP2 categories. Specificity was high for the Good (82%), Limited (98%) and Poor (95%) categories. PPV was high for Poor (82%) and NPV was high for all categories No differences were found between those with accurate or inaccurate predictions.

Conclusion: The results of the present study found that use of the PREP2 algorithm, with clinical measures only, is better than chance (chance = 25% for each of the 4 categories) alone at predicting a category of UL capacity at 3 months post stroke. PREP2 is a simple tool that facilitates evaluation of eventual UL outcome from measures routinely captured after a stroke within most healthcare settings in the US. The moderate to high values of sensitivity, specificity, PPV and NPV demonstrates the clinical utility of this predictive algorithm and it is recommended to be implemented into US clinical practice.

Neural and Behavioral Dissociation in Development of Coordination and Control Using a Prosthesis Simulator with Vibrotactile Feedback

John Johnson, Lewis Wheaton

Georgia Institute of Technology, Atlanta, USA

Rejection rates of upper-extremity prostheses remain high and clinical measures of their efficacy remain unchanged despite recent engineering advances in their construction. One technology purported to improve outcomes is augmented somatosensory feedback from the prosthesis. The question of whether augmented feedback from the prosthesis improves motor learning and control and the neural effects of this feedback remain open. We recruited healthy intact participants (n = 10) to use a prosthesis simulator to perform a reach-and-grasp task using three differently-sized discs. Vibrotactile feedback indicative of successfully grasping was supplied during the transport phase of the task during half of their trials. A second group of participants (n = 10) performed the same task using a pair of tongs without augmented feedback. Task phases of interest were reach-to-grasp and transport, using grasp aperture and velocity as measures of interest. Kinematic results during reach-to-grasp show a greater overshoot of grasp peak aperture for all disc sizes for prosthesis users compared to tongs users, with no effect of vibrotactile feedback. Participants using the prosthesis without feedback compared to with feedback demonstrated higher velocity during the initial 5%-15% of the reach-to-grasp phase, as well as during 10%-50% of the transport phase. The participants using the prosthesis in either condition (with or without feedback) demonstrated atypical velocity profiles during the reach-to-grasp phase, while their velocity profile during the transport phase was more stereotypical. Participants using the tongs showed a more efficient grasp peak aperture (less overshoot) than either prosthesis condition, as well, their velocity profiles for both reach-to-grasp and transport phases followed the bell-shaped velocity profile typical of human upper-extremity movements. Electroencephalographic data were analyzed on a trial-by-trial basis for three cortical regions of interest (frontal, left motor, left parietal) and two times of interest (grasp peak aperture and transport peak velocity) for alpha-band (10--14 Hz) power. At grasp peak aperture, results show that frontal midline and left motor activity were modulated by end-effector (prosthesis or tongs), as well as vibrotactile feedback, while left parietal activity was modulated solely by end-effector. At transport peak velocity, frontal and left parietal activity were modulated by end-effector. Left parietal activity was modulated by the presence or absence of feedback in the prosthesis group, while left motor only demonstrated a difference between tongs and prosthesis with feedback. Together, these differences in neural and behavioral results may suggest that participants establish coordination of muscle synergies during the time frame of the study, demonstrating an ability to develop synergies across the movement construction hierarchy, while control or scaling of movements remains under developed.

Characterizing the Impact of Baseline Cognitive Status on Dual Task Performance While Backward Protective Stepping

Jordan Barajas ¹ , Daniel Peterson ^1,2 , Linda Denney ³ , Shyamal Mehta ⁴

¹Arizona State University, Phoenix, USA. ²Phoenix VA Health Care Center, Phoenix, USA. ³Northern Arizona University, Phoenix, USA. ⁴Mayo Clinic, Scottsdale, USA

Background: The ability to concurrently perform cognitive and postural tasks (dual-tasking) relies on cognitive domains such as executive functioning and processing speed, which decline over time [1,2]. Further, these cognitive impairments are associated with decline in gait speed and increased fall risk [3]. A critical movement to avoid a fall after a loss of balance is a protective step [4]. We recently showed that protective stepping performance was not substantially impacted by a secondary cognitive task [5]. However, we observed large variability in cognitive capacity in our cohort. Characterizing the relationship between cognition and dual task ability (including prioritization across tasks) will provide additional insight into control of protective stepping under ecologically relevant (e.g., dual-tasking) scenarios. Thus, the purpose of this study was to determine the effect of baseline cognitive status on dual task interference and prioritization of postural and cognitive tasks while protective stepping.

Methods: 30 participants (Parkinson's disease (PD) n=16, healthy controls (HC) n=14) were divided into two groups based on their baseline cognitive status (measured by the Montreal Cognitive Assessment). Participants scoring <26 were sorted into the "low-cog group" (n=12) and those scoring ≥26 were placed in the "high-cog group" (n=18). Both groups experienced 2 cognitive trials where they verbally responded to an auditory Stroop test, 7 perturbation trials where they performed protective stepping after a quick, from-stance movement of the support surface (5 backward falls & 2 forward falls, randomly ordered), and 7 dual-task trials that combined the two tasks (5 backward, 2 forward, random order). Cognitive and protective stepping performance was calculated during single and dual tasking scenarios. The primary cognitive outcome variable was verbal reaction time and the primary stepping outcome variable was EMG onset of the tibialis anterior after backward protective steps.

Results: Contrary to our hypothesis, the low-cog group did not perform statistically significantly worse on any outcome measures compared to the high-cog group (p’s≥0.565). There was a group by task effect (p=0.028) for the step length dual task interference based on cognitive status, such that the low-cog group exhibited improvement in step length under dual task conditions compared to single task, but the high-cog group did not. For prioritization there was no statistically significant difference between the high-cog and low-cog groups, however the prioritization score for all outcomes was negative, indicating a stepping prioritization under dual task conditions for both groups.

Conclusions: This study provides preliminary evidence that baseline cognitive status does not significantly affect dual task interference nor prioritization while protective step dual tasking. While results should be treated with caution, they suggest that baseline cognitive status may not play a critical role in dual task interference or attentional allocation in both people with PD and healthy older adults.

The Role of Contralesional Motor Cortex During the Early Subacute Phase of Stroke Recovery

Kathleen Revill ¹ , Deborah Barany ² , Julie Tran ³ , Samir Belagaje ¹ , Fadi Nahab ¹ , Cathrin Buetefisch ¹

¹Emory University, Atlanta, USA. ²University of Georgia, Athens, USA. ³Emory Unversity, Atlanta, USA

After stroke, fMRI often reveals an increase in task-related contralesional (ipsilateral) primary motor cortex (M1) activation. While increased activity may indicate cortical reorganization after stroke, it could also be related to an increase in the difficulty of performing the motor task. Previous fMRI studies have shown that activation in ipsilateral M1 in healthy aged subjects increases as the demand for movement precision increases (Buetefisch et al. 2014, Barany et al. 2020). Demonstration of a strong relationship between activation of contralesional M1 in stroke patients and the demand for precision of a motor task would favor restoration of neurophysiological processes related to normal motor control. It is important to determine the relationship between contralesional M1 activity and stroke recovery as this area may serve as an additional target for neurorehabilitation treatment after stroke. Here we test the hypothesis that contralesional M1 activity is abnormally increased after stroke and is indicative of cortical reorganization rather than task demand. Main outcome measures were task-related BOLD signal in M1 and task accuracy. Stroke patients were also assessed for hand kinematics (velocity of wrist extension movements) and function (Jebsen Taylor Test).

In this study, patients (N = 19, 7M/12F, mean age = 59.4 years) with ischemic stroke affecting M1 output and corresponding paresis of the hand were scanned 1 month after stroke. Age- and sex-matched healthy controls (N = 31, 14M/17F, mean age = 61.7 years, all right-handed) were also scanned. Participants manipulated an MRI-compatible joystick to move a cursor into a target in one of four possible locations; stroke patients used the affected hand regardless of dominance and controls performed the task with both hands. Demand on precision was manipulated by varying target size across blocks. Electromyographic activity of the extensor carpi ulnaris muscle was monitored to ensure strictly unilateral hand movements. Participants also performed a visually paced hand flexion/extension task with both hands. Peak activity in this localizer task was used to define ipsilateral and contralateral regions of interest centered in M1 for each participant.

Preliminary results indicate that as target size increases, movement accuracy increases while movement time decreases, though stroke patients are less accurate and have longer movement times than controls. In keeping with previous reports, healthy control participants show demand-dependent activation in contralateral and ipsilateral M1, where increasing activation is seen with increasing demand for movement precision. Stroke patients also show demand-dependent activation in ipsilesional (contralateral) M1, but not contralesional M1. This suggests that abnormally increased contralesional M1 activity may additionally reflect reorganizational changes in these subacute stroke patients. Our ongoing work attempts to further substantiate this claim by relating these findings to other measures of hand function and kinematics, measures of M1 injury, and response to rTMS.

Brain-Computer Interface treatment for gait rehabilitation of chronic stroke patients

Marc Sebastián-Romagosa ¹ , Rupert Ortner ¹ , Woosang Cho ² , Katrin Mayr ² , Christoph Guger ^1,2

¹g.tec medical engineering, Barcelona, Spain. ²g.tec medical engineering, Schiedlberg, Austria

Brain-computer Interfaces (BCI) are used as a treatment for the motor rehabilitation in stroke survivors. The motor impairment that affects walking is one of the most frequent limitations, which causes a high functional deficit in the patient's autonomy. The objective of this work is to analyze the effectiveness of the BCI technology combined with Functional Electrical Stimulation for the motor rehabilitation of the lower extremity.

For this publication ten stroke patients in chronic phase with hemiparesis in the lower extremity were recruited. All of them participated in 25 BCI sessions about 3 times a week. The BCI system was based on the Motor Imagery (MI) of the paretic ankle dorsiflexion and healthy wrist dorsiflexion with Functional Electrical Stimulation (FES) and avatar feedback. Assessments were conducted to assess the changes in motor improvement before, after and during the rehabilitation training. Our primary measures used for the assessment were 10-meters walking test (10MWT), Range of Motion (ROM) of the ankle dorsiflexion and Timed Up and Go (TUG).

Results show a significant increase in the gait speed in the primary measure 10MWT fast velocity of 0.16 m/s (SD = 0.14). This improvement is above of the minimally clinically important difference (MCID). The speed in the TUG was also significantly increased by 0.06 m/s, P = 0.002. One patient was not able to perform TUG assessment before the rehabilitation training but was able to perform it after the BCI treatment with time 92.2 seconds. The passive ROM assessment increased 8.61° (SD = 6.54), P = 0.002, and active ROM increased 8.50° (SD = 7.23) after rehabilitation training, P =.008.

These outcomes show the feasibility of this BCI approach for chronic stroke patients, and further support the growing consensus that these types of tools might develop into a new paradigm for rehabilitation tool for stroke patients. However, the results are from only ten chronic stroke patients so the authors believe that this approach should be further validated in broader randomized controlled studies involving more patients.

MI and FES-based non-invasive BCIs are showing improvement for the gait rehabilitation of the patients in the chronic stage after stroke. This could have in impact on the rehabilitation techniques used for these patients, especially when they are severely impaired, and their mobility is limited.

Adapting telerehabilitation to COVID-19 regulations: preliminary results of participants’ satisfaction and effectiveness of an adapted telerehabilitation exercise program to improve upper limb function in individuals in the chronic phase of a stroke

Marie-Hélène Milot ^1,2 , Johanne Higgins ^3,4 , Hélène Corriveau ¹ , Louis-David Beaulieu ⁵ , Sonia Toy ⁶ , Marie-Hélène Boudrias ^6,7

¹Université de Sherbrooke, Sherbrooke, Canada. ²Centre de recherche sur le vieillissement, Sherbrooke, Canada. ³Université de Montréal, Montréal, Canada. ⁴CRIR, Montréal, Canada. ⁵Université du Québec à Chicoutimi, Chicoutimi, Canada. ⁶McGill University, Montreal, Canada. ⁷CRIR, Montreal, Canada

Background: Rehabilitation is effective in restoring function after a stroke. However, with the COVID-19 pandemic, many individuals with a stroke are no longer receiving adequate rehabilitation services. Telerehabilitation is a promising avenue to deliver at-home therapy. It was suggested that a hybrid model including a baseline in-person evaluation, is most effective and satisfying for the clinician and the patient but such hybrid model is currently limited or impossible to deliver due to the pandemic. Telerehabilitation must be adapted to respond to the COVID-19 reality and this pilot study proposes to do so by delivering telerehabilitation treatment without in-person initial assessment of participants.

Objectives: 1) Assess the feasibility (participants’ adherence and satisfaction) of using a COVID-19 adapted telerehabilitation intervention in chronic stroke survivors. 2) Explore the efficacy of the proposed telerehabilitation intervention at improving motor function of the affected upper limb (UL).

Methods: Nine participants (age: 58±15 years old, time since stroke: 97±66 months) participated in this non-randomized, single-group pilot study. Participants received a tablet to undergo a 6-week (3 times/week) telerehabilitation training program of their affected UL using the Physiotec application, a cloud-based bank of 15000 exercises, and Tera+, a telerehabilitation platform. Pre/post training, an evaluator performed a clinical evaluation, comprising sociodemographic data and UL motor function assessment (STREAM; normal =20), using the TERA+ platform. Training was adapted to each participant’s functional capacity, assessed during the baseline evaluation, and adjusted by a therapist at weeks #1-#3-#5, based on the participants progression. After completion of training, questionnaires on the participants’ satisfaction about the telerehabilitation and the exercise program as well as their perceived change of the UL were conducted. Wilcoxon signed ranked test was used to assess post-training changes in motor function.

Results: Out of 9 participants, one dropped out at the beginning of the study because the use of technology was too stressful. Following training, a significant change in STREAM score was observed (11.8±5.2 vs. 13.3±4.9 points; p= 0.015). A mean participants’ satisfaction rate of 96% with the use of telerehabilitation applications and exercises provided was noted with comments such as “Program went very well, really enjoyed it /There were some technical difficulties, but they were all resolved relatively easily”. Also, 6 out of 8 participants noted an improvement in the function of their UL following training, ranging from much to very much improved.

Conclusion: Telerehabilitation adapted to the COVID-19 reality is feasible and can translate into a significant improvement of the affected UL. Larger trials are needed to validate the preliminary results and assess if the proposed COVID-adapted telerehabilitation program can be suitable for other populations where containment has had a major impact on the delivery of care.

The incorporation of motor learning principles in virtual rehabilitation for individuals with cerebral palsy: a systematic review

Marika Demers ¹ , Karen Fung ² , Sandeep Subramanian ³ , Martin Lemay ^4,5 , Maxime Robert ^2,6

¹University of Southern California, Los Angeles, USA. ²Centre for Interdisciplinary Research in Rehabilitation and Social Integration, Quebec, Canada. ³UT Health San Antonio, San Antonio, USA. ⁴Centre de Recherche du CHU Sainte Justine, Montreal, Canada. ⁵Université du Québec à Montréal, Montreal, Canada. ⁶Université Laval, Quebec, Canada

Background: Virtual reality and active video games have unique attributes such as motivation, task difficulty, repetitive practice and enhanced feedback that make them an ideal modality to facilitate the incorporation of motor learning principles into the delivery of high intensity training for individuals with cerebral palsy. Increasing evidence supports the use of virtual reality and active video games to improve arm and hand function in children with cerebral palsy. However, it remains unclear if and how motor learning principles are incorporated in the development of rehabilitation interventions targeting arm and hand recovery using virtual reality.

Objective: To determine the extent to which motor learning principles are integrated in interventions using virtual reality and active video games to target upper limb function in individuals with cerebral palsy.

Methods: This systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Ten databases were searched using the key words and variations ‘cerebral palsy’, ‘virtual reality’, ‘video game’ and ‘rehabilitation’. Two members of the research team screened titles, Abstracts and full texts based on the inclusion and exclusion criteria. For all included studies, information about participants, study design, virtual reality system, dosage, type of practice, feedback provision and task-specificity were extracted. The quality of included studies was assessed with the modified Downs and Black scale.

Results: Thirty publications were included in this review. Most publications were classified as ‘fair’ on the modified Downs and Black checklist with a majority of small pilot or proof-of-concept studies. All publications included either school-aged children or adolescents. While our search strategy did not exclude studies involving adults with cerebral palsy, none of the studies included participants over 18 years old. Motor learning principles most frequently adopted were provision of enhanced feedback, variable practice, task-specificity and motivation. Dosage varied greatly between publications and only few publications reported the number of movement repetitions per session. Principles including progression of difficulty levels in games and assessment of retention and transfer of skills were poorly integrated, especially in commercial video game platforms and devices.

Conclusions: Virtual reality and active video games are well-suited to incorporate motor learning principles in rehabilitation interventions for children and adolescents with cerebral palsy. However, the ability to tailor the task difficulty to individual’s need remains to be better integrated in the development of future virtual reality systems for optimal arm and hand motor recovery in individuals with cerebral palsy. Future studies should assess how skills acquired through virtual reality and active video games are retained or transferred to real life activities to guide clinical decision-making.

Diffusion tensor-based morphometry detects volume loss in corticospinal tract associated with impaired motor recovery after stroke

Matthew Edwardson ^1,2 , Amritha Nayak ³ , Pooja Modi ⁴ , Neda Sadeghi ³ , Marie Luby ⁵ , Larry Latour ⁵ , Carlo Pierpaoli ³

¹Georgetown University, Washington, USA. ². ³National Institute of Brain Imaging and Bioengineering, NIH, Bethesda, USA. ⁴National Institute of Child Health and Human Development, NIH, Bethesda, USA. ⁵National Institute of Neurological Disorders and Stroke, NIH, Bethesda, USA

Background and Purpose: Wallerian degeneration in the corticospinal tract (CST) after stroke is characterized by decreased fractional anisotropy (FA) on MRI¹, which is associated with less stroke recovery^{2, 3} and response to therapy^{4, 5}. Diffusion tensor-based morphometry (D-TBM)⁶ is a recently proposed technique which enables the measurement of volumetric changes in brain structures from DTI data. We sought to determine whether D-TBM could detect volume loss in the CST and whether, like FA, these changes were associated with impaired stroke recovery.

Methods: We retrospectively analyzed clinical MRIs obtained from participants with acute ischemic stroke and unilateral arm impairment (NIHSS ≥ 2) and healthy controls presenting with TIA. Evolution maps for FA and the log of the determinant of the Jacobian (LnJ, representing volumetric change) were created by registering the 30-day image to the baseline (0-36hr) image and the resulting image to control space^{7, 8}. The evolution maps were separated into a good recovery group (Delta NIHSS arm motor item ≥ 2 from baseline to 30 days) and poor recovery group. Statistical maps identified clustered regions of interest (ROIs) with the greatest difference in FA and LnJ between the good and poor recovery groups. The association between arm recovery and FA and LnJ in the highest ranked ROIs was determined with Pearson correlations.

Results: Twenty-one participants (11 with good recovery and 10 with poor recovery) with anterior circulation stroke and 37 controls were evaluated. The evolution maps showed evidence of Wallerian degeneration that was more diffuse for FA extending into the cortical white matter and more focal for LnJ, localized to the brainstem and posterior limb of the internal capsule (PLIC). The statistical maps identified the top ranked ROIs for FA at the level of the PLIC, corona radiata, and cortical white matter whereas the best ROIs for LnJ were in the cerebral peduncle and pons. LnJ had a slightly stronger correlation with arm motor recovery than FA at the best performing ROI (R = 0.84, p = 1.8 x 10^-6 vs. R = -0.77, p = 4.9 x 10^-5 respectively).

Conclusions: D-TBM can detect Wallerian degeneration after stroke as evidenced by focal volume loss in the CST. These volumetric changes have a strong correlation with motor recovery and may provide complementary information to FA. Future studies will help determine whether D-TBM is useful for prediction of motor recovery and response to therapy.

Between vs. Within-Subject Predictors of Step Length Asymmetry Post-Stroke: One Predictor Does Not Fit All

Natalia Sanchez, Nicolas Schweighofer, James Finley

University of Southern California, Los Angeles, USA

Background: Gait impairment is common in the majority of survivors of stroke. A common measure of gait impairment post-stroke is step length asymmetry (SLA). SLA is associated with an increased energetic cost of walking¹, and decreased balance², making SLA reduction a common goal of clinical rehabilitation interventions^3,4. SLA is highly heterogeneous with some people walking with asymmetries characterized by longer paretic steps, others walking with asymmetries characterized by shorter paretic steps, and other individuals walking with nearly symmetric steps⁵. Studies assessing factors associated with SLA post-stroke traditionally select a subset of variables and obtain averages over multiple strides to understand between-subject differences in SLA. This approach reduces information by both eliminating within-subject variance and excluding most variables from analyses. We hypothesized that the between-subject relationships reported in the literature will not explain within-subject variance in SLA, and that these relationships depend on the direction of asymmetry.

Methods: Participants post-stroke walked on an instrumented treadmill at their self-selected speed. We recorded the position of reflective markers and ground reaction forces. We calculated spatiotemporal gait parameters, impulses, sagittal plane joint angles and sagittal plane joint moments adding up to 40 gait variables. We used sparse partial least square regression (sPLSR)^6,7, a technique that combines dimensionality reduction, variable selection, and multilevel regression to determine which of the 40 variables predicted within-subject variance in SLA in 19 individuals with chronic stroke.

Results: The predictors of within-subject variance in SLA differed from those reported to account for between-subject variance in SLA. The predictors of SLA also differed in participants who walked with shorter versus longer paretic steps. For participants who walked with longer paretic steps, 34% of the within-subjects variance in SLA was predicted by paretic double support time, braking impulse, peak vertical ground reaction force, and peak plantarflexion moment. Although peak paretic plantarflexion moment accounted for 4% of the within-subjects variance in SLA, it accounted for 42% of the between-subjects variance. In participants who walked with shorter paretic steps, paretic and non-paretic braking impulse predicted 28% of the within-subjects variance in SLA. Conversely, paretic braking impulse explained 68% of the between-subject variance in SLA, and the association between these two variables was in the opposite direction for within-subject vs. between-subject analyses.

Conclusion: We show that combined dimensionality reduction and regression can provide novel insights into the predictors of SLA post-stroke. This approach allows for explanation of between- and within-subjects variance in SLA without requiring a priori variable selection. Our findings indicate that the predictors that explain between-subject variance in SLA do not explain a within-subject variance in SLA. Overall, these results point to the need for developing approaches that take advantage of stride-to-stride variance, to identify personalized intervention targets for gait retraining.

Upper Limb and Trunk Responses to Repeated Lateral Perturbations in Aging

Nesreen Alissa ¹ , Ruth Akinlosotu ¹ , John D. Sorkin ^2,3 , George F. Wittenberg ^4,5 , Kelly Westlake ¹

¹Department of Physical Therapy and Rehabilitation Science, University of Maryland School of Medicine, Baltimore, Baltimore, USA. ²University of Maryland, Baltimore, Baltimore, USA. ³Baltimore VA Medical Center Geriatrics Research, Education, and Clinical Center, Baltimore, USA. ⁴VA Maryland HealthCare System, Department of Neurology, University of Maryland School of Medicine, Baltimore, Baltimore, USA. ⁵Department of Neurology, University of Pittsburgh School of Medicine, Pittsburgh, USA

Background: The upper limbs (UL) play an essential role in balance and fall prevention, especially in older adults (OA). Compared to young adults (YA), OA exhibit exaggerated first trial UL responses to a balance perturbation and delayed habituation over repeated trials. These exaggerated responses in OA may negatively impact balance recovery and fall prevention.

Objective: To examine age-related differences in UL and trunk responses to a first trial and habituation effects of a repeated lateral balance perturbation.

Methods: Eleven YA (25.8 ± 3.5 yrs) and 11 OA (70.3 ± 5.4 yrs) completed 10 left lateral perturbation trials using platform translations on the ActiveStep treadmill (Simbex, Inc.). Platform translations consisted of a quick 7.67 m/s² acceleration over 90 ms followed by a deceleration of 4.93 m/s² over 140 ms and a displacement of 0.14 m. Motion capture and electromyography were used to record UL and trunk movements and angles and activation patterns of the bilateral sternocleidomastoid (SCM), biceps, and middle deltoid muscles. Video recordings of the subjects’ feet captured stepping responses. In-trial falls were determined using a load cell attached to the harness. The Activities-Specific Balance Confidence (ABC) Scale was also administered.

Results: OA had greater maximum shoulder abduction (left p < 0.0001; right p < 0.0001) and elbow flexion (left p = 0.001; right p < 0.0001) across trials 1 and 10. OA had greater left lateral trunk flexion in trial 1 (p = 0.002), but not trial 10. OA also had greater shoulder abduction (left p = 0.0002; right p < 0.0001), right elbow flexion (p = 0.005), and left lateral trunk flexion (p = 0.002) at first foot lift off (FFLO) in trial 1. No differences were found in muscle onset latencies. However, there were age-related differences in onset latency of FFLO (p = 0.03) in trials 1 and 10. There were age-related differences in time to peak EMG amplitude for the left SCM (p = 0.01), left middle deltoid (p = 0.03), and biceps (left p = 0.02; right p = 0.01). Habituation occurred earlier in YA compared to OA with no evidence of habituation in left shoulder abduction and right elbow flexion in OA. OA tended to narrow their base of support (BOS) whereas YA widened their BOS. There was also a negative correlation between ABC scores and right shoulder abduction (r = −0.639, p < 0.05) among OA during trial 1.

Conclusions: First trial responses are exaggerated, and habituation is delayed in OA compared to YA. There are age-related differences in step type which result in reduced BOS in OA and increased BOS in YA. Age-related differences in shoulder and trunk angles at FFLO suggest that UL response may influence step type.

Real-Time Gait and Posture Measures from Wearable Sensors for Rehabilitation in Parkinson’s Disease

Niveditha Muthukrishnan ¹ , Abhi Ashwinkumar Mevawala ¹ , Holly A. Shill ² , James J. Abbas ¹ , Narayanan Krishnamurthi ³

¹Arizona State University, Tempe, USA. ²Muhammad Ali Parkinson Center, Phoenix, USA. ³Arizona State University, Phoenix, USA

Gait and posture impairments in Parkinson’s disease (PD) significantly affect quality of life and lead to an increased risk of falls. These impairments are usually manifested as reduced step/stride length, increased asymmetry and variability, and stooped posture; they worsen with disease progression and are resistant to pharmacological and surgical treatments. Cue-based rehabilitation techniques have shown promise for improving gait and posture symptoms. However, cues delivered in a preset manner do not present information about the user’s current performance, and therefore have limited effectiveness in modulating performance. This limitation can be addressed by providing real-time feedback on the user’s performance. In prior work, we demonstrated improvements in gait and posture by providing real-time visual feedback of step length and back angle during treadmill walking in people with mild to moderate PD [1,2]. Participants were able to utilize the feedback successfully and improve their gait patterns and posture. This study also demonstrated that people with PD could follow feedback on a step-by-step basis. However, this treadmill-based visual feedback was restricted to a laboratory setting and required sophisticated instrumentation. The goals of the current study are to design and develop effective rehabilitation strategies using wearable sensors integrated with a mobile application for real-time measurement and feedback of gait and posture measures. Such a system could enable feedback-based rehabilitation outside of a laboratory setting, i.e., at home or in the user’s community. The specific aims of this project are to systematically develop a real-time feedback system using wearable sensors that will: 1) measure step length, step time, and back angle in real-time, and 2) provide feedback of computed measures in real-time, with the intent to improve gait and stooped posture through enhanced movement awareness. As the first step in this process, we have developed an algorithm that can process data streamed from wearable sensors to calculate step-based gait measures. The developed algorithm implements a direct integration method with additional processing such as a coordinate transformation and drift-correction. In an evaluation of subjects with and without PD, the step length and step time measures computed using this algorithm were in close agreement with the gold-standard system [3]. In on-going work, we are developing algorithms to calculate back angle measures and developing real-time versions of these algorithms. The computed real-time gait and back angle measures will be compared with simultaneous measures obtained from a laboratory-based gold-standard system. Agreement analysis will be conducted for the measures computed using Passing-Bablok regression, Intraclass Correlation (absolute agreement), and Bland-Altman plots. If the real-time wearable measurement system is demonstrated to be accurate and reliable, it will then be used to provide feedback for gait and posture rehabilitation in PD.

Understanding Intracortical Excitability in Phantom Limb Pain: A Multivariate Analysis from a Multicenter Randomized Clinical Trial

Paulo Teixeira ^1,2 , Kevin Pacheco-Barrios ^1,2 , Muhammed Enes Gunduz ^1,2 , Anna C. Gianlourenco ^1,2 , Luis Castelo-Branco ^1,2 , Felipe Fregni ¹

¹Harvard Medical School, Boston, USA. ²Spaulding Rehabilitation Hospital, Boston, USA

The perpetuation of pain has been associated with maladaptive neuroplasticity [1, 2, 3]. Past research has focused on the cortical dysfunction and maladaptive neuroplasticity in people with chronic pain and suggested that altered intracortical excitability contributes to the chronicity of pain perception [4]. Cortex excitability and its modulatory interactions are suggested to have a key role with sensory areas in chronic pain subjects. However, the findings on the characteristics of cortex excitability in individuals with chronic pain are contradictory. To explore the associations of intracortical excitability with clinical characteristics in a large sample of subjects with chronic pain can contribute to understanding how cortex excitability behaves in chronic pain conditions and serves as a starting point to better use this neurophysiological marker as a biomarker of pain.

Objective: To explore associations of intracortical excitability with clinical characteristics in a large sample of subjects with phantom limb pain (PLP).

Methods: Ancillary study using baseline and longitudinal data from a large multicenter randomized trial that investigated the effects of non-invasive brain stimulation combined with sensorimotor training on PLP. Multivariate regression modeling analyses were used to investigate the association of intracortical excitability, measured by percentages of intracortical inhibition (ICI) and facilitation (ICF) with clinical variables.

Results: Ninety-eight subjects were included. Phantom sensation of itching was positively associated with ICI changes and at baseline in the affected hemisphere (contralateral to PLP). However, in the non-affected hemisphere (ipsilateral to PLP), the phantom sensation of warmth and PLP intensity were negatively associated with ICI (both models). For the ICF, PLP intensity (baseline model only) and age (longitudinal model) were negatively associated while time since amputation and amputation level (both for longitudinal model only) were positively associated in the affected hemisphere. Additionally, use of antidepressants led to lower ICF in the non-affected hemisphere for the baseline model while higher amputation level also led to less changes in the ICF.

Conclusion: Results revealed clear associations of clinical variables and cortex excitability in a large chronic pain sample. ICI and ICF changes are not to be mainly explained by PLP intensity. Instead, other variables associated with duration of neuroplasticity changes (such as age and duration of amputation) and compensatory mechanisms (such as itching and phantom limb sensation) seem to be more important to explain these variables.

What is the scientific premise of current therapeutic approaches in motor rehabilitation after stroke?: A thematic analysis

Rajiv Ranganathan ¹ , Carson Doherty ¹ , Michael Gussert ¹ , Eva Kaplinski ¹ , Mary Koje ² , Chandramouli Krishnan ²

¹Michigan State University, East Lansing, USA. ²University of Michigan, Ann Arbor, USA

Despite tremendous advances in treatment and management of stroke, restoring motor and functional outcomes after stroke continues to be a major clinical challenge. Given the wide range of approaches used in motor rehabilitation, recent commentaries have highlighted the lack of a clear scientific premise for different interventions as one factor that has led to sub-optimal study outcomes. To examine this issue in greater detail, we conducted a thematic analysis of randomized controlled trials in stroke rehabilitation over the last 2 years to better understand the scientific premise of current therapeutic approaches.

A PubMed search was conducted with search terms “stroke” and “rehabilitation” with filters enabled for “clinical trial” and “randomized control trial”. The extracted articles were manually screened to identify articles that met the following inclusion criteria: (1) prospective randomized clinical trial, (2) outcomes were primarily related to motor aspect of stroke rehabilitation, (3) presence of a ‘usual care’, conventional, or sham control group, and (4) articles written in English. Studies that used invasive approaches, drug/pharmacological interventions, or alternative/complementary medicinal approaches (e.g., acupuncture, TaiChi) were excluded.

From an initial search of 539 articles, 99 articles satisfied our inclusion criteria. From these studies, we examined the stated premise/rationale for the intervention. We also extracted other parameters such as the age, time since stroke, sample size, and whether the intervention outperformed the control group in the primary outcome.

About 50% of the studies showed an effect where the intervention group outperformed the control group. Our results revealed three primary findings – (i) a majority of studies either did not explicitly provide a rationale or simply pointed to the results of a prior study as the basis of their work. (ii) when a rationale was stated, the majority of approaches mentioned increasing intensity of therapy, increasing enjoyment and motivation, and/or increasing the access to therapy (low cost/home rehab, etc.), and (iii) scientific premises based on basic science or current theories of motor learning were rare.

These results highlight the need for strengthening cross-disciplinary connections between basic science and clinical studies. Further development and testing of the scientific premise behind different approaches to rehabilitation is critical for developing effective treatment interventions.

Insights into Coordination Deficits After Stroke: A Refined Analysis of the Lower Extremity Motor Coordination Test (LEMOCOT)

Shirley Handelzalts ^1,2 , Yogev Koren ^1,2 , Noy Goldhammer ² , Adi Yeshurun ² , Simona Bar-Haim ^1,2

¹Ben-Gurion University, Beer Sheva, Israel. ²The Translational Neurorehabilitation Lab at Adi Negev Nahalat Eran, Ofakim, Israel

Background: Coordination of lower extremity movements is often impaired in persons with stroke (PwS). In the lower extremity motor coordination test (LEMOCOT) the participant is asked to alternately move the big toe between a proximal and a distal target (30 cm apart) as many times as possible in 20 seconds. The total number of ‘in target’ touches constitutes the score.^1-3 In this study we investigate whether a more refined analysis of the performance in the LEMOCOT can provide further insights into the underlying deficit after stroke. This might contribute to our understanding of recovery processes after stroke. We aimed to characterize the LEMOCOT performance in PwS and healthy controls by means of performance accuracy.

Methods: Twenty PwS (62±12 years; 84±83 days after stroke onset; lower extremity Fugl-Meyer score 30.2±3.7) and 20 healthy controls (42±16 years) participated in this cross-sectional study. Participants performed the test in a seated position while moving the foot between targets marked on an electronic mat, equipped with force sensors with a sampling rate of 60 Hz (Zebris FDM-T Treadmill, Zebris Medical GmbH, Germany). A dedicated algorithm and a software were developed to extract the contact area of each touch, from which the following parameters were computed: center of the area (COA), center of pressure (COP) and the distance between them. In addition, the absolute and variable error were calculated for each touch. For statistical analysis we used mixed-effect models with ‘participants’ as the random effect. For all dependent variables we used full factorial models with ‘group’ (stroke/controls), ‘leg’ (paretic/non paretic/controls) and ‘target’ (proximal/distal) as main effects.

Results: For all outcome measures we found a significant main effect for the ‘group’, indicating superior performance for the control group. In addition to the effects observed for the number of ‘in target’ touches, we found significant group by target interactions for contact area, absolute error and variable error, indicating that the main effect of ‘group’ was derived from the performance in the distal target. A significant main effect for ‘leg’ and a group by leg interaction were found for variable error, indicating that the main effect of ‘leg’ was derived from an increased error in the paretic lower extremity. Also, a significant group by leg by target interaction was found for the distance between the COA and COP.

Conclusions: Findings indicate that PwS preformed worse than controls in all outcome measures, primarily due to their difficulty to accurately touch the distal target. Several of our parameters such as the distance between COA and COP and variable error, were able to detect differences in performance accuracy that are not captured by the conventional parameter of ‘in target’ counts.

Myoelectric upper limb orthosis use in motor learning therapy in chronic stroke

Svetlana Pundik ^1,2 , Jessica McCabe ¹ , Margaret Skelly ¹ , Ahlam Salameh ^1,2 , Zhengyi Chen ² , Curtis Tatsuoka ² , Stefania Fatone ³

¹VA NothEast Ohio HealthCare system, Cleveland, USA. ²Case Western Reserve Univeristy, Cleveland, USA. ³Northwestern University, Chicago, USA

Introduction: Arm deficits are common after stroke with recovery often limited. Adjuvant treatments that enhance motor learning-based therapy may improve outcome. The MyoPro® is a myoelectric controlled upper limb orthosis that harnesses the user’s electromyography to assist the user to move the weak arm. Our purpose was to assess the feasibility of delivering combination motor learning-based therapy and MyoPro in chronic stroke.

Methods: Seven subjects (28.6% male; 3.7±3.1 years post-stroke) with severe upper limb impairment participated in an In-clinic phase (2x/wk for 9 weeks) followed by a Home phase (9 weeks). In-clinic phase included 27 hours of face to face therapy with and without the MyoPro and home exercise program (HEP). During the Home phase, participants continued their HEP only. Therapy was based on motor-learning principles of timely progression and high repetition of close to normal movement. Outcomes were Fugl-Meyer (FM), Modified Ashworth (MAS; sum of 9 muscles), Chedoke Arm and Hand Activity Inventory (CAHAI), Craig Handicap Assessment and Rehabilitation Tool (CHART), and Orthotic and Prosthetic Users’ Survey satisfaction module (OPUSsat). Outcomes were collected at baseline, after In-clinic, and after Home phases. Longitudinal linear mixed effect models included fixed effects for time points, baseline/age adjustment, and random effects of subjects to account for within-subject correlation.

Results: Baseline characteristics included: (mean(SE)): FM=24.3(2.1), MAS=8.5(1.1), CAHAI=20.0(2.2), CHART=321.8(35.1) and OPUSsat=62.2(8.0). Subjects performed 7229.0 (2056.0) hand reps and 2428.9 (878.7) with MyoPro across both phases. Statistically significant changes after treatment and controlling for baseline were observed for all outcomes. Significant changes from baseline to end of In-clinic phase and from baseline to end of Home phase, respectively, were: FM improvement of 6.0(1.2) (p<0.0001) and 6.7(1.3) (p<0.0001); MAS improvement of 2.1(0.4) (p<0.0001) and 1.79(0.46) (p=0.0004); CAHAI improvement of 5.7(0.96) (p<0.0001) and 8.1(0.96) (p<0.0001); and OPUSsat improvement of 15.8(4.6) (p=0.03) and 15.2(4.63) (p=0.03). CHART improved at the end of Home phase 55.49(14.46) (p=0.0086) but not at end of In-clinic phase 23.06(14.46) (p=0.16).

Conclusions: Use of MyoPro in motor-learning-based therapy was feasible and produced clinically meaningful gains in response to a therapy program with relatively short face to face duration on a measure of motor function (FM; MCID 5.25 points). Gains in impairment (FM, MAS), function (CAHAI), satisfaction (OPUSsat) and participation (CHART) were observed regardless of baseline level. This upper limb therapeutic approach may allow high quality practice of movement for individuals with severe motor deficits after stroke.

Metabolome changes are seen in people with Multiple Sclerosis who respond to progressive resistance training

Jennifer Keller ¹ , Pavan Bhargava ² , Kathleen Zackowski ^3,2

¹Kennedy Krieger Institute, Baltimore, USA. ²Johns Hopkins University, Baltimore, USA. ³National Multiple Sclerosis Society, New York, USA

Background: People with multiple sclerosis show variable responsiveness to progressive resistance training. The mechanisms by which exercise produces beneficial effects in MS are not well understood and adaptations to metabolic pathways may underlie the differences between responders and non-responders.

Objective: The goal of this study was to evaluate the effects of lower extremity progressive resistance training (PRT) on the metabolome of people with MS (pwMS) and evaluate the extent that metabolomic changes are related to exercise responsiveness.

Methods: 14 pwMS with EDSS < 4.0 and 13 age- and sex-matched healthy controls completed a 12-week, in-person PRT exercise intervention. Outcome measures included: Expanded Disability Status Scale (EDSS), fatigue (modified fatigue impact scale [MFIS]), hip strength, walking speed (timed 25-foot walk [T25FW], walking distance (six-minute walk test-[6MWT], cardiovascular fitness tests, and plasma metabolomics analysis. PRT responders were identified by a change in hip strength over a minimal detectable change. Nonparametric statistics (significance set at p<0.05) were used to assess for between group differences (Kruskal-Wallis test) and within group changes pre-post training (Wicoxon matched-pair signed rank test). Spearman’s Rank-Order correlations were used to determine relationships between measures that showed change.

Results: Both MS and control groups, improved hip strength, walking distance, and walking speed, by contrast fatigue was improved only in pwMS. Fatty acid, phospholipid and sex steroid metabolism were significantly altered by PRT in pwMS. Change in these metabolite modules as well as in fatigue and walking outcomes were significantly different between responders (N=8) and non-responders (N=6) for pwMS.

Conclusion: Participants with MS who showed improvements in strength and fatigue also showed changes in fatty acid, phospholipid and sex steroid metabolism. A larger confirmatory study is needed to further investigate these findings and explore the opportunity of metabolic supplementation. Supplementation may have dual benefits to augment exercise intervention responsiveness and to address fatigue.

The Current Landscape of Exercise and Physical Rehabilitation Services for People Living With Multiple Sclerosis in Saskatchewan: A Formal Environmental Scan and Needs Assessment

Kristen Plandowski ¹ , Cameron Mang ¹ , Sarah Donkers ²

¹Faculty of Kinesiology and Health Studies, University of Regina, Regina, Canada. ²College of Medicine, University of Saskatchewan, Saskatoon, Canada

Background: Exercise is beneficial for people living with multiple sclerosis (PwMS) both in terms of physical function and overall quality of life. However, many PwMS are not meeting the recommended guidelines for physical activity. This is concerning given the cumulative functional impairments typically experienced over the multiple sclerosis (MS) disease course. MS is the most common neurological condition experienced by young adults. Saskatchewan (SK), Canada has one of the highest rates of MS in the world. Through our recent work, we have determined that PwMS in SK want access to programs and services to support them with their physical activity behaviour, but find that there are few options available for them.

Objective: Our overall objective is to identify and describe the existing exercise and physical rehabilitation programs and services for PwMS in SK. To achieve this objective, we aim to: 1) conduct a formal environmental scan of available programs and services, 2) characterize the regional distribution, components and specificity of each, and 3) determine the key needs of service providers related to the implementation of such programs and services.

Methods: This study employs a multi-phase patient-centred approach to systematically capture the current SK landscape and needs with regards to exercise and physical rehabilitation programs and services for PwMS. The first step was a systematic web-based search to identify available programs and services located within SK. Using predetermined keywords and inclusion/exclusion criteria, a broad, exhaustive internet search was conducted from August to October, 2020. Identified programs and services and available details were extracted into a data collection table and categorized according to population size, provider type, and type of physical intervention. These findings, paired with prior knowledge from authors’ programs of research, were used to develop an online survey to further identify and explore existing services and programs. Results from the survey will be used to elaborate upon the initial search findings and identify the key needs of service providers. To ensure a comprehensive and robust process, preliminary results will be shared with an expert review panel of key stakeholders to verify our findings and interpretations, and to plan next steps.

Results: The initial search identified 25 relevant services, all in urban areas. Approximately 75% are located in cities with a population size ≥ 100,000. Less than 25% were MS-specific. Providers ranged from rehabilitation professionals offering one-on-one therapy (36%) to fitness instructors providing in-person group exercise classes (44%). We are awaiting survey results, which will be presented at the Annual Meeting.

Conclusion: Preliminary findings suggest limited program availability, particularly in rural areas. Although PwMS in SK on disease modifying therapies have on-going access to nursing support, professional support for addressing physical activity behaviour is desired but not widely available.

Spontaneous neural synchrony links intrinsic spinal sensory and motor networks during unconsciousness

Jacob McPherson, Maria Bandres

Washington University School of Medicine, St. Louis, USA

Purposeful functional connectivity during unconsciousness is a defining feature of supraspinal networks. However, its generalizability to intrinsic spinal networks remains incompletely understood. The most reliable findings have been correlations between spontaneous BOLD signals in the left and right dorsal horns, and, separately, the left and right ventral horns. Spontaneous connectivity between the dorsal and ventral horns, between the intermediate gray and the ventral horn, and within the ventral horn itself have yet to be reliably delineated. It is also not readily apparent whether structured intraspinal connectivity persists at the single-unit level during unconsciousness and in the absence of evoked neural transmission.

The potential function(s) of resting state intraspinal connectivity are likewise unknown. An intriguing possibility is that it plays a role in adaptive or maladaptive neural plasticity through a form of reactivation and synaptic stabilization during unconsciousness. This hypothesis is drawn from the function of supraspinal network activity during sleep, and is supported by the finding of altered patterns of BOLD-based intraspinal functional connectivity in conditions associated with maladaptive neural plasticity in spinal networks. To have a direct role in shaping neural plasticity, however, a necessary substrate would be the tandem presence of synchronous discharge amongst populations of individual units spanning multiple spatial and functional regions.

Given the importance of the spinal cord for sensorimotor integration and reflexes, we reasoned that spontaneous functional connectivity between neurons in sensory-dominant and motor-dominant regions of the gray matter would be a precondition for purposeful network activity during unconsciousness. And for the reasons noted above, such a finding would have important implications for both the physiological and pathophysiological states. Here, we address three unresolved questions. First, is neuron-level functional connectivity evident in regions of the spinal gray matter not traditionally associated with primary afferent inflow? Second, is spontaneous functional connectivity evident between sensory and motor regions of the gray matter? And third, does the proportion of spontaneously active neurons exhibiting correlated discharge, as well as their topology, depart from that which would be expected amongst an interconnected population of statistically similar neurons firing uncooperatively (i.e., randomly)?

We addressed these questions in vivo in rats (N=22), recording large populations of single units throughout the dorso-ventral extent of the lumbar enlargement. We find that robust spontaneous neural activity is prevalent throughout the gray matter during unconsciousness and that neurons in sensory and motor regions exhibit significant, non-random correlations in their spatiotemporal discharge patterns. This functional connectivity could not be explained by latent afferent activity or by populations of interconnected neurons spiking uncooperatively. The observed activity is consistent with a network policy in which salient, experience-dependent patterns of neural transmission introduced during behavior or by injury/disease are reactivated during unconsciousness.

Vagus Nerve Stimulation Paired with Rehabilitation to Improve Recovery of Upper Extremity Function in Stroke and Spinal Cord Injury

Seth Hays ^1,2 , Chad Swank ³ , Robert Rennaker ^1,2 , Jennifer French ⁴

¹Texas Biomedical Device Center, Dallas, USA. ²University of Texas at Dallas, Dallas, USA. ³Baylor Scott & White Institute for Rehabilitation, Dallas, USA. ⁴Neurotech Network, St. Petersburg, USA

Vagus nerve stimulation (VNS) drives precisely-timed release of norepinephrine and acetylcholine, which are known to function synergistically to govern synaptic plasticity. Preclinical studies demonstrate that vagus nerve stimulation (VNS) paired with sensorimotor rehabilitative training significantly improves recovery in a variety of animal models of neurological disorders, including spinal cord injury (SCI), stroke, traumatic brain injury, and peripheral nerve injury. Mechanistic studies in rats reveal that VNS-dependent benefits require the activation of neuromodulatory systems and arise from synaptic plasticity in central networks. Recently completed Phase 3 pivotal study demonstrated that coupling vagus nerve stimulation (VNS)-dependent release of pro-plasticity neuromodulators with rehabilitation yielded a clinically significant three-fold increase in upper extremity motor function compared to rehabilitation without VNS in people with chronic stroke, highlighting the clinical translatability of this Targeted Plasticity Therapy (TPT). These findings position VNS paired with rehabilitation to be the first evidence-based neuromodulation intervention for chronic stroke. We will present results from the pre-clinical studies using an implantable VNS and external interface technology with SCI and stroke models demonstrating significant improvements of upper extremity function for the VNS + Rehabilitation group over the control group. These studies provide a mechanism of action basis for this novel VNS technology. Clinical translational efforts will also be highlighted.

Characterization of spontaneous sensorimotor neural transmission in the adult spinal cord in vivo

Maria F Bandres, Jacob G McPherson

Washington University School on Medicine, St. Louis, USA

Spontaneous action potential discharge (spAP) is a ubiquitous and functionally meaningful feature of the developing spinal cord. It generates the complex patterns of neural transmission that are necessary to link sensory inputs with motor outputs and to shape nascent behaviors. However, the extent to which purposeful, spatiotemporally diverse spAP persists in the adult spinal cord is unclear. Thus, its presence remains a central question in understanding experience-dependent modification of behavior and (mal)adaptative responses to injury or disease.

Reports of the location and prevalence of spAP in the adult spinal cord have typically been confined to restricted anatomical regions of the gray matter or to specific types of networks or processes (e.g., nociceptive, motor). Thus, its overall distribution remains unresolved and its relative prevalence across anatomical regions cannot be readily inferred from the available literature. Here, we attempted to reconcile some of the outstanding contradictions that emerge from this generally disintegrated picture of adult spinal spAP. Specifically, we recorded simultaneously from large populations of interneurons distributed throughout the dorso-ventral extent of the gray matter in vivo and used spike train analyses to quantify spatiotemporal features of spAP.

All experiments were approved by the Institutional Animal Care and Usage Committees of Florida International University and Washington University in St. Louis. Experiments were conducted in adult male Sprague-Dawley rats under urethane anesthesia (N=11). After T13-L2 laminectomy, microelectrode arrays were implanted at the L5 dorsal root entry zone. Electrode locations for neuronal recording spanned the dorsal horn, intermediate gray, and ventral horn. Primary outcome measures included the number of spontaneously active neurons; temporal features of spAP spike trains, such as mean frequency, regularity, and burstiness; and the spatiotemporal profile of neural activity throughout the gray matter – i.e., potential variations in temporal features of spAP by anatomical region.

We found robust and spatiotemporally diverse spAP throughout the extent of the sampled gray matter. The majority of cells (>80%) presented irregular firing patterns (including bursting and non-bursting subgroups). Irregular firing patterns are associated with high levels of spike train variability in neurons. Within each anatomical region, spike train variability also delineated multiple subgroups of neurons. Interestingly, mean firing frequency, inter-burst frequency, and intra-burst frequency all increased monotonically moving ventrally from the dorsal surface of the cord, while spike train variability remained unchanged.

Widespread concurrence of spAP in sensory, motor, and integrative regions, coupled with high spike train variability, would be predicted during salient neural transmission in intrinsic spinal networks. Our findings are consistent with this framework. Thus, it is possible that spAP may maintain a purposeful role even in mature, non-developing spinal networks. Future work will be required to determine if this is indeed the case, and, if so, to elucidate its potential function(s).

Potentially Preventable Readmissions after Inpatient Rehabilitation

Amanda Herrmann ^1,2 , Ella Chrenka ^1,2 , Leah Hanson ^1,2 , Haitham Hussein ^1,2,3 , Gretchen Niemioja ^1,3

¹HealthPartners Neuroscience Center, St Paul, USA. ²HealthPartners Institute, Minneapolis, USA. ³Regions Hospital, St Paul, USA

Background: Centers for Medicare and Medicaid Services (CMS) developed quality measures for tracking unplanned, potentially preventable readmissions (PPRs) within 90 days after discharge from an inpatient rehabilitation facility (IRF). The purpose of this study was to examine the trend for PPRs in patients who have discharged from our hospital’s IRF over the last several years.

Methods: For this study, we retrospectively identified individuals who discharged from our hospital’s acute IRF and were readmitted to our hospital with a PPR diagnosis within 90 days after discharge. We exported patient data from an internal hospital database and patients’ electronic health records. A control group of age and sex matched patients, who discharged from the IRF but did not experience a PPR was randomly selected from the same database. Univariate analyses were used to compare patient demographics, medical histories, and functional conditions between the two groups. Logistic regression estimated the adjusted odds of experiencing a PPR associated with race, primary diagnosis, length of stay for initial IRF stay, admission Functional Independence Measure (FIM), and count of medications, comorbidities, complications, and safety events.

Results: Between 2013 and 2019, 75 patients experienced a 90 day PPR (2.9% of all patients). The average age was 63 ± 15 years and 64% of the sample was male. The two groups did not significantly differ in race and ethnicity with most patients being white and non-Hispanic (White: 75%, Non-Hispanic: 94%). A primary diagnosis of stroke was most common (PPR: 47%, control: 51%), however the PPR group was more likely to include patients with spinal cord injury (23%, 9%). The average time to readmission was 32 ± 24 days. Sepsis was the most common cause of PPR (33%) followed by renal failure (16%). FIM Motor Scores were significantly lower in the PPR group at the time of admission (p<0.01) and discharge (p=0.03). PPR patients had significantly higher numbers of comorbidities as compared to the control group (5.9 per patient, 4 per patient, p<0.01). While patients in both group had comparable numbers of medications (2.6, 2.3, p=0.13), hypnotics were significantly more common in the PPR group (21%, 5%, p<0.01) and anticoagulants were more common in the control group (64%, 88%,p<0.01). Logistic regression estimated a 47% increase in odds of PPR associated with one additional comorbidity [Adjusted OR 1.47 (1.23, 1.77), p <.01]. None of the other 6 variables showed an independent association with PPR.

Conclusion: In patients who discharged from our hospital’s IRF, the most common PPR diagnoses were sepsis and renal failure. PPR patients also had greater medical complexity, as demonstrated by a larger number of comorbidities and lower FIM Motor Scores on IRF admission and discharge. Practitioners should be vigilant in patients who meet these criteria.

Virtual Lab Toolbox: Infrastructure and reliability of translating in-lab motor tasks to an online, unsupervised version

Andrew Hooyman, Sydney Schaefer

Arizona State University, Tempe, USA

The capability to continue human movement research during an unprecedented era of limited face-to-face contact is imperative. Fortunately, converting some established lab-based human movement paradigms to an online space is possible at minimal cost using common internet tools and open-source software (i.e., a virtual lab). However, navigating technical limitations within this space and determining the reliability is critical. The purpose of this Abstract is to 1) outline the infrastructure needed to translate an experimental motor paradigm from a typical face-to-face laboratory setting into a secure online space, and 2) demonstrate the reliability of motor data collected online. Motor tasks that require button presses only are ideal for translation to an online space using a free software known as Unity. We have found that Unity, a gaming software, is capable of reliably collecting and storing continuous cursor position and acceleration data at 100 Hz. To deploy online, the motor task was converted into a Unity game and hosted on an online server as a web graphic library (WebGL). Currently, prices for secure server and domain name space cost approximately 500 US dollars for a 4-year commitment. This comes with 3 GB of storage space within a relational database that is capable of storing all standard forms of data, i.e. numeric, integer and character as both short and very long vectors. Unity WebGL objects can be directed to have their data projected and stored within the relational database. Then, to determine the reliability of this approach, we played a previously-developed motor learning video game requiring sequential key presses (SuperG) and the data were stored directly to the computer hard drive and the online relational database simultaneously to compare the two data storage methods directly. The data stored online was 99.8% identical to that of the data stored locally, thereby demonstrating that the proposed methods are reliable and affordable for online, unsupervised collection of (certain types) of human movement data.

Virtual Lab Toolbox: Application of machine learning for grading figure-drawing visuospatial tests

Andrew Hooyman, Jessica Trevino, Sydney Schaefer

Arizona State University, Tempe, USA

Visuospatial tests such as the Rey-Osterrieth Complex Figure Test (ROCFT) require the drawing of a complex figure that is then graded by a trained neuropsychologist. Recent research has demonstrated that scores on this assessment are linked to motor skill learning as well as cognitive status (e.g., Mild Cognitive Impairment diagnosis). With the advent of blood-based biomarkers and other telehealth tools (developed both prior to and during the COVID-19 pandemic), collecting and grading the ROCF and other related measures (e.g., clock drawing) remotely and/or at a large scale remain challenging. We therefore present here a web-based analytical pipeline that automates the scoring of hand-drawn figure tests such as the ROCFT. To test this pipeline, we first compiled a dataset of 134 ROCFTs from a group of cognitively intact older adults (N = 45, mean age: 65.5 yo; 29 F) that were graded manually per the test instructions and verified by a clinical neuropsychologist. The drawings were then scanned to generate individual images of each assessment and then decomposed into pixel space. Next, a support vector machine regression machine learning algorithm was trained on 80% of the complete dataset. The model’s capability to predict the correct grade was then tested on the remaining 20% of the data. Results demonstrated that the machine learning algorithm was comparable to the true grade +/- 5 points (out of 36), yielding an ICC of a two-way random effects model = .77. Additional use of this tool is expected to improve the prediction of the algorithm by expanding the size of the training set. This tool not only allows for fast and easy assessment of complex figure drawing, but also enables the evaluation of individuals remotely.

Visual Recognition Algorithm in Differential Diagnosis Between Glioblastoma and Solitary Brain Metastasis in Patients With No History of Systemic Cancer: A Mri-Based Study

Camilla Russo ¹ , Francesca Lettieri ¹ , Angelo Russo ¹ , Guido Maria Secondulfo ¹ , Fernanda Picozzi ¹ , Alfredo Marinelli ² , Paolo Maresca ¹

¹Department of Electrical Engineering and Information Technology (DIETI), University of Naples “Federico II”, Naples, Italy. ²IRCCS Neuromed Istituto Neurologico Mediterraneo Pozzilli (INM), Pozzilli, Italy

Background: Brain metastases (MET), secondary malignant localizations at a distance from a primary site of cancer, represent the most common central nervous system (CNS) malignancy in adults. Primary malignant brain tumors arising from brain tissue are far less common; among them glioblastoma (GBM) is the most aggressive one, accounting for more than half of all primary malignant CNS lesions. Non-invasive early differentiation of single brain MET from GBM is crucial to tumor staging, prognosis prediction and timely therapeutic intervention. Indeed extra-neural spreading of GBM is an exceptional occurrence; therefore in these cases systemic imaging is not required. Conversely, in patients with suspected single brain MET without previous history of cancer, systemic examination is imperative for staging and targeted treatment. IBM Watson™ Visual Recognition system is an artificial intelligence (AI) tool, already used as prototype for pattern recognition in precision oncology. In this preliminary retrospective study, we tested the application of the IBM Watson™ for Magnetic Resonance Imaging (MRI) pattern recognition od GBM vs solitary brain MET, highlighting possible decision-support implications.

Methods: In this preliminary study, we analysed axial T1-weighted 3D gradient-echo contrast-enhanced MR sequences acquired 10’ after bolus injection of gadolinium-based contrast agent (Gadobutrol, 0.1mmol/kg), on 1.5T MRI scans from three different vendors. All patients were first diagnosed with brain tumor and did not receive any medical/surgical treatment before imaging; all MRI examinations were performed for surgical navigation purposes between 2018 and 2020. Overall 58 GBM and 26 single brain MET were retrospectively analysed, with a total number of 710 key-images from GBM and 246 key-images from single brain MET finally included. Custom models were defined on IBM Watson™ Visual Recognition system to classify lesions in two groups: GBM vs single brain MET. Images preparation started with preliminary conversion of DICOM images in supported files format (.JPG); then images were randomly divided into training data set (70%) and validation data set (30%). The model performance was analysed in terms of accuracy, precision, recall & F1 score metrics.

Results and Conclusions: Training and testing cycle was carried on by the IBM Watson™ Visual Recognition deep learning algorithm in 1h4’28’’. Global classification accuracy was 0,77. Results were 0.78 precision, 0.92 recall, 0.85 F1 score for GBM, and 0.72 precision, 0.45 recall, 0.56 F1 score for MET. Our results support a possible role for AI classifiers in improving decision-making in neuro-oncology. IBM Watson™ Visual Recognition deep learning algorithm seems to provide an accurate early differentiation between GBM and solitary MET, with higher performances for GBM probably due to the larger training set provided. In this light, iterative training on larger MR images set, as well as on different MRI sequences and/or acquisition planes is still required to confirm these preliminary findings.

Volumetric Assessment of White Matter Lesion Burden and Grey Matter Volume Changes in Mild Cognitive Impairment: A Clustering Analysis

Camilla Russo ¹ , Elisa Capone ¹ , Luigi Della Gatta ¹ , Alfredo Di Gaeta ¹ , Francesco Laquaniti ¹ , Eugenio Capobianco ¹ , Eduardo Gragnano ² , Rossana Senese ³ , Mario Muto ¹

¹Department of Neuroradiology, A.O.R.N. Cardarelli, Naples, Italy. ²Department of Precision Medicine, University of Campania “Luigi Vanvitelli”, Naples, Italy. ³Emicenter European Medical Imaging, Via Taverna Rossa, 169, 80020, Casavatore, Naples, Italy

Purpose: Dementia is a syndromic disease characterized by memory loss, decline in intellectual abilities, and deterioration in cognitive/behavioural functioning, with devastating impact on everyday activities; although more common in older people, dementia is not a physiologic manifestation of normal ageing. Mild cognitive impairment (MCI) is generally an early manifestation of dementia, defined as intellectual decline greater than expected for age/education, but still not interfering with daily life. A prompt identification of MCI signs and symptoms is important for a timely treatment, in order to slow down disease progression to dementia and provide appropriate caregivers’ support. Frequently post-mortem brain examination represents the only tool for definite differential diagnosis between different types of dementia; however structural magnetic resonance imaging (MRI) can provide important clues in supporting early clinical diagnosis of both vascular cognitive impairment (VCI), primary neurodegenerative dementias (PND), and dual pathologies (DP) where vascular damage and primary neurodegeneration both contribute to brain atrophy. In this pilot study, we investigated whether the relation of white matter abnormalities and brain volume changes at MRI may help in distinguishing between different pathological mechanisms beyond MCI before dementia onset.

Materials: From June to December 2020 we selected 46 age- and sex-matched individuals (mean age 53,9±19,3y; 17M:34F), 33 patients with clinical diagnosis of MCI and 13 healthy controls (HC); among the MCI patients, 13 were clinically suspected with predominantly vascular dementia and 20 with possible primary neurodegeneration. All subjects underwent brain MRI on the same 3T scanner, and 3D-T1w and 3D-FLAIR sequences were acquired. The medical software Icobrain^dm (version 4.3) was used for automated correction for head size and volume segmentation, both of white matter hyperintensities on 3D-FLAIR and grey matter volumes on 3D-T1w images. Finally cluster analysis on multimodal MRI data was performed, and similarity between groups was measured by Ward’s clustering linkage method; after choosing the number of clusters (4), significant differences among subgroups of patients and controls were evaluated using analysis of variance. Statistical analysis was performed using Xlstat software (version 2019.7).

Results: K-means cluster analysis clearly divided our sample in four clusters (13HC, 13PND, 9VCI and 11DP, respectively); subsequent hierarchical dendrogram showed a large distance between chosen cut-off and the Ward’s similarity axis, suggesting that the resulting groups are quite different from each other. When performing the analysis of variance, p-value is <0.05, thus supporting the evidence of significant different means among the four clusters.

Conclusions: Volumetric structural MRI can provide important information for MCI management, helping in early differential diagnosis between VCI, PND and DP event before overt dementia onset. However further studies on larger samples coupled to longitudinal patients’ evaluation are imperatively required to assess the reliability of cluster analysis in this specific clinical setting.

Transcranial Magnetic Stimulation (TMS) Analysis Toolbox: A user friendly open source Matlab GUI for basic and advanced analysis of TMS related outcomes

David Cunningham ^1,2 , Bei Zhang ^1,2 , Aaron Cahn ^1,2

¹Case Western Reserve University, Cleveland, USA. ²MetroHealth Systems, Cleveland, USA

Single and paired-pulse transcranial magnetic stimulation (TMS) is used in the field of neurorehabilitation in order to investigate adaptive/maladaptive neuroplasticity as a result of neurological injury and recovery. TMS uses electromagnetic induction to depolarize upper motor neurons which results in propagation of action potentials to the targeted muscle. The resulting compound muscle action potential (motor evoked potential [MEP]) and suppression of muscle activity (silent periods) are recorded with electromyography (EMG) at the targeted muscle. Often, the EMG signal can be noisy depending on the patient population, which makes it difficult to analyze the outcome of interest without visualizing the data. Further, basic and advanced analysis require knowledge of programming in order to efficiently and reliably analyze large datasets. For example, it is necessary to perform data reduction based on background EMG activation (active vs. rest) and to group trials for input/output curve analysis or paired-pulse protocols. We have developed TMS Analysis ToolBox, a user friendly Matlab program with a graphical user interface that can perform basic and advanced analysis of common TMS related outcomes on individual or averaged TMS trials (e.g. MEP latency/amplitudes, silent periods (duration and % decrease), input/output curves (sigmoidal fitting and area under the curve) and EMG onset detection. The toolbox imports whole multi-channel files and time-locks the data based on comments, data blocks, or thresholds (e.g. TTL). Further, it allows interactive analysis for data reduction and outcome detection for immediate visualization and exporting of results for second level analyses. TMS analysis toolbox currently supports file exports from: LabChart, Brain Vision, AcqKnowledge and Signal, and is in beta mode with full functionality. Our intent is to include more analysis functions, make it open source and widely accessible. For more information and/or to provide data from alternate data acquisition software for inclusion, please contact: David Cunningham, PhD ([e-mail]).

Neurorehabilitative gaming across the continuum of care during the COVID-19 pandemic and beyond

Hayley Haaf ¹ , Harrison Segall ¹ , Ken Johnson ¹ , Steve Zeiler ² , Amelia Tenberg ² , Mona Bahouth ² , Preeti Raghavan ^3,1

¹Department of Physical Medicine and Rehabilitation, Baltimore, USA. ²Department of Neurology, Baltimore, USA. ³Johns Hopkins University School of Medicine, Baltimore, USA

Objectives: Providing high dose rehabilitation remains a challenge during and after hospitalization; the COVID-19 crisis has worsened this problem. Technology-enhanced rehabilitation can increase access to training activities in multiple settings. The MindMotion™GO (MMGO), an FDA-approved mobile rehabilitation platform developed by neuroscientists to gamify rehabilitation, is one such technology. Here we describe the clinical deployment of the MMGO to provide rehabilitation services during COVID-19.

Methods: The MMGO consists of a Microsoft Kinect to track large movements and a Leap Motion sensor to track fine finger, wrist and forearm movements, which connects to a computer with 26 games. The games are calibrated to the individual, and target strength, dexterity, endurance, balance, coordination, trunk flexibility, and gait. In the hospital, a therapist used the MMGO synchronously in addition to providing standard care and facilitated social distancing where possible. Once the patient left the hospital, a therapist confirmed calibration of the MMGO in the patient’s home during synchronous telerehabilitation visits. An asynchronous rehabilitation protocol facilitated a home-exercise program and continuity of rehabilitation. The MMGO companion website tracked the usage of game play and provided data to modify game difficulty and/or duration remotely in real-time during routine telerehabilitation visits.

Results: From March through July 2020, 82 patients received treatment with the MMGO: 27 patients utilized it in the outpatient setting, 26 received it during the inpatient hospital stay, and 29 received it during inpatient rehabilitation. Of the 27 patients who utilized the device as outpatients, 19 of them received a unit at home for both synchronous and asynchronous use. The MMGO facilitated therapists’ ability to evaluate and treat patients remotely despite the COVID-19 restrictions. Duration of use for patients ranged from several days up to 8 weeks, depending on the setting of care.

Conclusions: The clinical deployment of MMGO is feasible and enabled accessibility to rehabilitation across the continuum of care during the COVID-19 pandemic. Technology-enhanced telerehabilitation has the potential to augment rehabilitation services even after COVID-19 pandemic restrictions are lifted. Further study of this model of care is ongoing.

Expanding Horizons of Rehabilitation Technology with 3D Printable Elastics: A Case Series of Stroke Survivors

Justin Huber

University of Kentucky, Lexington, KY, USA

In neurorehabilitation, wearable devices serve many important roles in patient care—therapeutic intervention (e.g. biofeedback sensors and functional electrical stimulation), bracing (e.g. anti-spasticity orthoses), and functional restoration (e.g. tenodesis splints and soft robotics). A growing demand for rehabilitation devices is prompting new approaches to the fabrication of wearable technology. Additive fabrication with 3D printing offers many benefits including unique geometries, rapid prototypes, and personalized devices. Research on 3D printable orthotic devices has primarily focused on rigid constructs such as arm and ankle braces. This study investigates non-rigid, 3D printable elastic materials and explores their application to a dynamic orthosis. A dynamic hand orthosis was designed to feature an interchangeable finger component. This component of the orthosis was duplicated in five 3D printed elastic materials. By interchanging this finger component, each elastic material underwent usability testing by volunteer participants. In a case series, two participants with chronic hand impairment secondary to stroke performed the usability tests. Outcomes included qualitative participant feedback and quantitative metrics—pincer force, pincer aperture, and the Box and Block Test. This study demonstrates 3D printable elastic materials in a dynamic orthosis, which may inspire future applications in soft robotics and wearable technologies.

Key words. Dynamic Orthosis, Additive Manufacturing, Assistive Technology, Hemiparesis, Upper Extremity, Hand

OnabotulinumtoxinA Treatment in Patients with Upper Limb and Lower Limb Spasticity from the ASPIRE Study

Ganesh Bavikatte ¹ , Gerard Francisco ² , Alberto Esquenazi ³ , Michael Dimyan ⁴ , Kenneth Ngo ⁵ , Marc Schwartz ⁶ , Aleksej Zuzek ⁷ , Wolfgang Jost ^8,9

¹The Walton Centre, Liverpool, United Kingdom. ²University of Texas McGovern Medical School and TIRR Memorial Hermann, Houston, USA. ³MossRehab Gait and Motion Analysis Laboratory, Elkins Park, USA. ⁴University of Maryland, School of Medicine, Baltimore, USA. ⁵Brooks Rehabilitation Hospital, Jacksonville, USA. ⁶MS Biostatistics, LLC, Clermont, USA. ⁷Allergan, an AbbVie company, Irvine, USA. ⁸Parkinson-Klinik Ortenau, Germany. ⁹University of Freiburg, Department of Neurology, Wolfach, Germany

Objective: Examine onabotulinumtoxinA utilization in patients with upper limb (UL) and lower limb (LL) spasticity from the Adult Spasticity International Registry (ASPIRE) study to gain real-world insights into the treatment of spastic hemiparesis.

Design: 2-year, international, multicenter, prospective, observational registry (NCT01930786). Adult patients with spasticity across multiple etiologies were treated with onabotulinumtoxinA at the clinician’s discretion. OnabotulinumtoxinA utilization and safety data were collected at each treatment session. Patients with spastic hemiparesis were defined as receiving ≥1 UL treatment and ≥1 LL treatment during the study.

Results: ASPIRE patients (N=730) were on average 53.6 years old, 52% female, 77% Caucasian, 37% naïve to botulinum toxin(s) for spasticity, and 56% post-stroke. Of N=730, n=284 patients were defined as hemiparetic. In hemiparetic patients treated for UL and LL at the same session (n=275), the mean total dose of onabotulinumtoxinA was 477 U for UL+LL, 257 U for UL, and 220 U for LL. Of n=275 hemiparetic patients, 56% had a treatment interval of 10-15 weeks, 62% had 5-15 injections/session, and 82% had >5 muscles injected/session. Clenched fist was the most common UL presentation (n=219 patients), with 55% of sessions to the left side only. Equinovarus foot was the most common LL presentation (n=238 patients), with 52% of sessions to the left side only. Of the hemiparetic population (n=284), 115 patients (41%) reported 375 adverse events; 42 patients (15%) reported 80 serious adverse events. No new safety signals were identified.

Conclusions: This preliminary analysis from ASPIRE provides valuable, real-world evidence on the use of onabotulinumtoxinA to treat patients with combined upper limb and lower limb spasticity. OnabotulinumtoxinA was most frequently utilized to treat clenched fist (UL) and equinovarus foot (LL) in patients with spastic hemiparesis.

An objective hand proprioception assessment system for pediatric and adult population

Jinseok Oh ¹ , Arash Mahnan ¹ , Jessica Holst-Wolf ¹ , Jiapeng Xu ¹ , Hannah Block ² , Juergen Konczak ¹

¹University of Minnesota, Minneapolis, USA. ²Indiana University Bloomington, Bloomington, USA

Proprioception is the perception of joint and body position and movement in space. The sense is essential to motor control and learning such as grabbing a key or learning to type with a keyboard. Several neurological disorders, such as Parkinson's disease and stroke, have been associated with the deprivation of the sense. The same is true for several pediatric disorders such as developmental coordination disorder or cerebral palsy. Current clinical practice to measure the loss of proprioception, nonetheless, relies on the subjective reports of the patients. For example, clinicians may bend a patient's index finger at two different angles and ask if the two positions were different. Quantifying the amount the patients' deficit would benefit the clinicians in two aspects. It would help them prescribe the right amount of rehabilitation exercises. It would also let them better evaluate how effective their prescriptions were. Consequently, several attempts have been made to quantify the degree of proprioceptive dysfunction. Many of them, however, require complex robotic devices that are expensive and need skilled personnel to control. This paper addresses the use of an assessment system that comprises a tablet and an accompanying software application for objectively assessing hand proprioception. Users place their hands underneath a stand that holds a tablet. The tablet occludes the vision of the hand. While the index finger is stationary, the tablet screen presents two color-coded areas, and users have to judge, in which of the two areas their unseen index finger is located. For each trial, the system receives a users' response and implements a Bayesian-based adaptive psychophysical threshold hunting method. The method updates the values of the variables required for the display configuration. After approximately 10 minutes of testing, the system calculates the user's perceptual bias, defined as the difference between the point of subjective equality (PSE), the position of the user's assumed finger position, and the actual finger position. Seven healthy adults (age = 22.4 ± 4.4, F: 3) were tested with the system. The mean perceptual bias was 2.0° (sd = 4.2°). The system demonstrates a decent ability to measure the proprioceptive acuity with a precision of 0.1°. The interactive feature and simplicity of our system distinguish it from conventional proprioceptive assessment methods that passively displace a limb or limb segment. The proposed system provides an engaging testing environment for young children who have a short attention span, or people with impaired movement with reduced cognitive ability (e.g. due to cortical stroke or Parkinson's disease). Lastly, the system is highly adaptive and accommodates tablets of different sizes and users with diverse hand sizes. Thus, the system has the potential to be used as an assessment tool for evaluating proprioceptive deficits in children and adults.

Vibro-tactile stimulation as a non-invasive neuromodulation approach for cervical dystonia: a case study

Yi Zhu, Arash Mahnan, Jürgen Konczak

University of Minnesota, Minneapolis, USA

Introduction: Cervical dystonia (CD) is a type of focal dystonia characterized by involuntary neck postures¹. The underlying neurophysiology mechanism of CD is unknown, but there is increasing empirical evidence that motor deficits of CD are associated with somatosensory and proprioceptive deficits^2,3. Vibro-tactile stimulation (VTS) is a non-invasive somatosensory stimulation approach where afferent signals from the vibrated muscle and tactile mechanoreceptors modulate cortical activity⁴. Previous studies have shown that VTS could be an effective neuromodulation therapy for treating laryngeal dystonia⁵. This proof-of-concept study examined the effect of VTS on alleviating the involuntary cervical muscle contractions in two adolescents with different presentations of CD – a male with consistent retrocollis and a female with intermittent torticollis.

Method: VTS was applied sequentially on four neck positions: bilateral trapezius (TRP) and bilateral sternocleidomastoid (SCM). Each VTS site was vibrated continuously for six minutes. The kinematics and underlying neck muscle activities during dystonic neck movements were examined with acceleration and surface electromyography (sEMG). Two acceleration features and two sEMG features were derived: (1) number of peaks per minute; (2) peak amplitude of acceleration (PAA); (3) change in power of sEMG after VTS; (4) cumulative density function of sEMG between 3-10 Hz (CDF₁₀).

Results: First, the application of neck muscle VTS did not induce meaningful symptom relief for the participant with constant retrocollis. Second, the frequency of dystonic neck movements decreased by 60% after VTS in the participant who presented with intermittent torticollis. In addition, PAA during dystonic episodes was significantly lower in post VTS when compared to baseline. Third, the effectiveness of VTS in alleviating dystonic muscle spasms depended on the site of vibration. For the patient with right torticollis, the left trapezius muscle was the optimal vibration site reducing sEMG signal power by 15% across all recorded muscles. During VTS on the optimal vibration site, sEMG power of left trapezius dropped rapidly within the first minute and then continued to decline over time. As an additional result, the mean CDF₁₀ of left trapezius in post VTS condition was found significantly lower than baseline.

Discussion: This case study offered preliminary insight into the assumed effectiveness of neck muscle VTS as a treatment for CD. One participant responded positively to VTS. The frequency and extent of the dystonic postures were markedly reduced during and immediately after VTS application. A systematic study with a larger sample size is required in the future to validate the effectiveness of VTS for treating symptoms in CD.

Aerobic exercise and neuroplasticity in spinal cord injury: a systematic review

Anjali Sivaramakrishnan¹, Aditi Hombali²

¹UT Health San Antonio, San Antonio, USA. ²

Background: There is emerging evidence that exercise can induce neuroplasticity in the central nervous system following a spinal cord injury (SCI). Intact, non-damaged tracts and inter-neuronal circuits after an incomplete SCI have the capacity to reorganize and adapt spontaneously and in response to training. Animal studies in SCI have shown that aerobic exercise (AE) induced change in neurotrophin levels, circuit reorganization and neuronal morphology, however evidence in humans is limited. Although some studies suggest that biomarkers of neuroplasticity may upregulate with AE and improve spinal plasticity, the relationship between exercise parameters and neuroplasticity in SCI is unclear. In this systematic review we synthesized the literature for understanding the effects of aerobic exercise on neuroplasticity in individuals with SCI.

Methods: A systematic search of electronic databases including OVID (Medline), CINAHL and EMBASE was conducted with keywords related to SCI, neuroplasticity and AE. We included randomized and non-randomized studies involving adults with a clinical diagnosis of spinal cord injury; utilizing AE as an intervention including arm ergometry, walking, cycling, treadmill training alone or in combination with another intervention; and administering at least one neuroplasticity-based outcome (neurotrophic growth factors, imaging measures, evoked potentials, spinal reflex excitability measures). Animal studies and single case studies were excluded from this review.

Results: Our synthesized findings suggest that long term exercise may normalize spinal reflex excitability, improve cortical excitability and cortical activation. The effects of AE on neurotrophin levels are inconsistent, where some acute AE studies showed an exercise-dependent increase in brain derived neurotrophic growth factor (BDNF) while others reported no change.

Conclusion: This review provides evidence that AE may induce structural and functional neuroplasticity in SCI. However, there is marked heterogeneity in study design, participant characteristics, exercise intervention and reported neuroplasticity outcomes. While it is apparent that functional motor recovery may occur in parallel with changes in the corticospinal drive to the spinal and peripheral circuits, more research is warranted to identify the optimal exercise parameters and the relationship to neuroplasticity mechanisms. Future studies are also required to better delineate the role of exercise intensity on neurotrophin levels.

Neuroplasticity underlying balance recovery in individuals with incomplete SCI: Effects of a strength and perturbation-based training

Charlotte H. Pion ^1,2,3 , Dorothy Barthélemy ^1,2,3,4

¹École de réadaptation, Faculté de Médecine, Université de Montréal, Montreal, Canada. ²Centre de Recherche Interdisciplinaire en Réadaptation du Montréal métropolitain (CRIR), Montreal, Canada. ³Institut universitaire sur la réadaptation en déficience physique de Montréal (IURDPM), Montreal, Canada. ⁴Research Center, Hôpital du Sacré-Coeur de Montréal, Montreal, Canada

Introduction: Following incomplete Spinal Cord Injury (iSCI), stability during standing and gait is poor, and the risk of falls is increased. Postural reactions, which consist of specific muscle activation enabling balance recovery following a perturbation, are delayed and decreased post-iSCI which underlies such increased fall risk. However, deficits in postural reactions are not fully understood post-iSCi and are not routinely assessed clinically. Moreover, therapeutical approaches to improve these postural reactions have not been established.

Objective of this pilot study: Determine the impact of a 4-week combined strength and perturbation-based training on postural reactions and balance control in individuals with iSCI and identify neural mechanisms underlying these improvements.

Methods: Two individuals with chronic iSCI (1 man; 41±14 years old; ASIA D) were assessed 1 week prior, 1 week after and 1 month after a training combining explosive ankle strength and standing balance training (4 weeks; 3x/week; 1h15/session). Assessment comprised: 1) Measures of postural reactions including analysis of center of pressure (CoP; displacement in mediolateral and anteroposterior axis; 95% ellipse area and total excursion) and pattern of Electromyographic activation following backward and forward perturbations; 2) the Mini-Best test. Measures to identify neuroplastic mechanisms were: 1) Sensorimotor integration at cortical level using the short-latency afferent inhibition in soleus (tibial nerve: 1.5xmotor threshold (MT); Transcranial magnetic stimulation (TMS);1.1xMT; interstimulus interval (ISI):25-100ms); 2) Intracortical excitability by paired pulse TMS: test:1.1xMT; conditioning:0.7xMT; ISI=2ms (inhibition) and 12ms (facilitation) and 3) Modulation of the soleus H-reflex by the corticospinal tract (subthreshold TMS=0.9xMT; Tibial nerve stimulation= 5% of maximal M-wave; ISI=0-20ms).

Results: Improved postural reactions: Prior to training, the 95% confidence area was larger and onset latency of TA activation was delayed during backward tilt in iSCI participants compared to controls. Following training, the 95% confidence area decreased by 20% and onset latency of TA activation following perturbation was closer to the control values. However, the Mini-Best test remained unchanged following training.

Neuroplasticity. Improvement in postural reactions was paralleled by changes in neuroplasticity measures. The short-latency afferent inhibition (ISI=60 ms) decreased after training leading to increased MEP amplitude. Intracortical facilitation increased following training, also leading to greater MEP amplitude. Changes in the corticospinal modulation of the H-reflex were not clear as the facilitation of H-reflex typically observed at those delays was increased in one participant and decreased in the other iSCI participant. At 1-month post-training, the improvements observed in postural reactions were partially maintained but neuroplastic changes were no longer observed.

Conclusion: These preliminary results suggest that combined strength and perturbation-based training led to improved postural reactions by enhancing sensorimotor integration and cortical excitability. Although behavioral improvements are still partially observed 1-month post-training, neuroplastic changes are not maintained in the pathways assessed.

Use of Computational Modeling to Inform tDCS Electrode Montages for the Promotion of Motor Recovery in Spinal Cord Injury

Maria Martin ^1,2 , Ileana Therese Mendoza ^1,2 , Kelsey Baker ²

¹Department of Health and Biomedical Sciences, College of Health Professions, UTRGV, Edinburg, USA. ²Department of Molecular Sciences, UTRGV School of Medicine, Edinburg, USA

Objective: Following a spinal cord injury (SCI), damage in the spinal cord can hinder neuronal signals sent from the brain to the periphery resulting in varying degrees of paralysis and loss of function. We have found that one way to boost hindered signals is by pairing rehabilitation with transcranial direct current stimulation (tDCS). However, it remains unclear how the focality of tDCS to the targeted region affects functional outcomes. Therefore, our objective was to determine if tDCS focality influenced motor outcomes in individuals with SCI.

Methods: Twelve subjects with cervical SCI were enrolled and received rehabilitation for two weeks. During rehabilitation, we applied conventional tDCS (7x5 cm electrode) to the site of a weaker muscle based on the location of their SCI. Here, the tDCS anode was applied to the motor cortex at the site of the weaker muscle and the cathode was placed on the contralateral supraorbital area. We assessed changes in motor function before and after rehabilitation using the upper extremity motor score (UEMS), nine hole peg test (NHPT) and action research arm test (ARAT). We assessed how well tDCS current targeted the weaker muscle site using the software HD-Explore. We analyzed the relationship between the amount of current applied to the cortical representation and changes in motor function after rehabilitation directly using SPSS.

Results: We found that the amount of current targeting the weak muscle site ranged from 0.313 to 0.528 V/m. We did not see a correlation between the amount of tDCS current and the change in motor function UEMS (R=0.1356), proximal strength (R=0.1309), total strength (R=0.1275), NHPT (R=0.0032), ARAT (R=0.0562). Our initial work suggests that the amount of current being delivered to the site of the weaker muscle does not influence upper limb motor function in patients with chronic spinal cord injury. Current work is evaluating how current intensity delivered to the stronger muscles may also influence upper limb motor recovery in patients with chronic spinal cord injury.

Conclusion: Transcranial direct current stimulation (tDCS) is a noninvasive form of brain stimulation that delivers current to targeted areas of the brain. However, it remains unclear how the focality of tDCS to the targeted region affects functional outcomes. Here we evaluated how tDCS focality influenced motor outcomes in individuals with SCI. Overall, our findings suggest that directly targeting current to the cortical representation of the weak muscle does not improve upper limb functional motor recovery in chronic spinal cord injury patients. Current work is evaluating the relationship between current intensity delivered to the strong muscles that may also influence upper limb functional recovery in patients with chronic spinal cord injury.

Effect of remote ischemic conditioning on hand function in cervical spinal cord injury

William Savage ¹ , Yu-Kuang Wu ^1,2 , Grace Famodimu ¹ , Gregory Mendez ¹ , Noam Harel ^2,3

¹Bronx Veterans Medical Research Foundation, Bronx, USA. ²Icahn School of Medicine at Mount Sinai, New York, USA. ³James J. Peters VA Medical Center, Bronx, USA

Background: Most spinal cord injuries (SCI) are not full transections, indicating that residual nerve circuits are retained after injury. Activity-based SCI rehabilitation interventions, including physical training and neural stimulation, have been shown to beneficially reorganize spared motor pathways in the brain, corticospinal tract, and at the spinal level. The need remains, however, for an effective approach to synergistically increase the magnitude and duration of neuroplasticity induced by other interventions. Remote ischemic conditioning (RIC) has been shown to promote neural plasticity while requiring minimal equipment (a timer and a manual blood pressure device). When one limb is exposed to several cycles of transient ischemia, a systemic (“remote”) protective effect occurs across a range of tissues subjected to subsequent injury. RIC may also mediate an activating effect on neural circuits. For this reason, we are investigating RIC coupled with physical training to promote neuroplasticity in hand muscles after cervical SCI. This is the first study introducing RIC in the SCI population. We hypothesize that RIC will acutely synergize with motor task training via increasing corticospinal excitability.

Methods: This is a prospective randomized-order crossover trial. Participants undergo two experimental sessions consisting of 50 minutes of either active or sham RIC preceding two minutes of hand exercise. The primary outcome is motor evoked potentials (MEPs) of the target abductor pollicis brevis (APB) muscle at four time points: baseline, post-RIC, post-exercise, and 15 minutes post-exercise. The post-RIC measurement allows us to first observe whether RIC alone acutely affects corticospinal excitability. The post-exercise measurements allow us to investigate if RIC synergizes with exercise. Secondary outcomes include effects on gene expression of neurotrophic factors and inflammatory mediators, as well as effects on cardiovascular parameters.

Results: Due to Covid-related disruptions, only three participants have completed the experiments to date. In two able-bodied participants, peak-to-peak MEP amplitude at 120% of resting motor threshold (MEP₁₂₀) increased by 36.8% following active-RIC plus exercise and decreased by 31.6% following sham-RIC plus exercise. In one SCI participant (a 55-year-old male with chronic C4 level injury, ASIA impairment scale C), MEP₁₂₀ increased by 30.2% following active-RIC plus exercise and increased by 42.4% following sham-RIC plus exercise. Gene expression and cardiovascular exploratory outcomes have not been analyzed yet.

Discussion: We are actively enrolling additional SCI and uninjured participants to better evaluate the effect of RIC and motor training on corticospinal excitability. If RIC is found to improve rehabilitation from SCI, widespread use would be feasible. Our future goal would be to couple RIC with more prolonged rehabilitation training to promote long-term effects.

Repetitive peripheral sensory stimulation as an add-on intervention for upper limb rehabilitation in stroke: a randomized trial

Adriana Conforto ¹ , André Machado ^2,3 , Nathalia Ribeiro ¹ , Ela Plow ^2,3 , Sook-Lei Liew ⁴ , Claudia Leite ⁵ , Artemis Zavaliangos-Petropulu ⁴ , Isabella Menezes ¹ , Sarah Anjos ⁶ , Rafael Luccas ¹ , Leonardo Cohen ⁷

¹Hospital das Clínicas, São Paulo University, São Paulo, Brazil. ²Cleveland Clinic, Cleveland, USA. ³Case Western Reserve University, Cleveland, USA. ⁴University of Southern California, Los Angeles, USA. ⁵São Paulo University, São Paulo, Brazil. ⁶University of Alabama, Birmingham, USA. ⁷NINDS, NIH, Bethesda, USA

Introduction: Repetitive peripheral sensory stimulation (RPSS) followed by 4-hour task-specific training improves upper limb motor function in subjects with stroke who experience moderate to severe motor upper limb impairments. Here, we compared effects of RPSS and sham followed by a shorter duration of training, in subjects with moderate to severe motor impairments in the chronic phase after stroke.

Methods: This single-center, randomized, placebo-controlled, parallel-group clinical trial compared effects of 18 sessions of either 1.5 hour active RPSS or sham followed by supervised 30-minute functional electrical stimulation (FES) and 45-minute task-oriented training (TOT) of the paretic upper limb. In both groups, subjects were instructed to perform functional tasks at home. The primary outcome measure was the Wolf Motor Function Test (WMFT-TIME and functional assessment scale, WMFT-FAS) after six weeks of treatment. Grasp and pinch strength were secondary outcomes.

Results: In intention-to-treat analysis, WMFT-TIME improved significantly in both active and sham groups at 3 and 6 weeks of treatment. Grasp strength improved significantly in the active, but not in the sham group, at 3 and 6 weeks. Pinch strength improved significantly in both groups at 3, and only in the active group at 6 weeks.

Conclusions: RPSS enhanced hand strength in chronic stroke. Despite the short duration of supervised training, changes in WMFT-TIME in both groups and in WMFT-FAS in the active group were comparable to those reported after longer, more intensive training protocols. These findings are relevant to settings that impose constraints in duration of direct contact between therapists and patients.

Probing sensorimotor integration within the spinal cord using paired, non-invasive stimulation in post-stroke and neurotypical older adults

Alejandro Lopez ¹ , Jiang Xu ² , Lena Ting ¹ , Michael Borich ¹ , Trisha Kesar ¹

¹Emory University, Atlanta, USA. ²Tongji Hospital, Wuhan, China

Stroke causes disruption in the orchestrated integration of sensorimotor neural circuits implicated in the control of gait and balance. However, aging-related declines can also contribute to impaired sensorimotor circuit function post-stroke. Given that the prevalence of stroke increases with age, our study aims to parse out aging-related declines and neurologic impairment within spinal sensorimotor circuits. Here, we utilize paired cortical and peripheral nerve stimulation (PNS) to evaluate timing- and activity-dependent of descending modulation of spinal segmental reflexes. In chronic stroke survivors versus neurotypical older adults (NOA), we aimed to compare (1) descending modulation of spinal segmental reflexes, and (2) activity- and timing-dependent influences on descending modulation of spinal segmental reflexes. To date, 7 NOA (6 females; 50-79 years) and 5 chronic stroke survivors (mild-to-moderate impairment; 3 females; 46-68 years) have been evaluated. PNS was delivered to the posterior tibial nerve to elicit soleus Hoffmann (H-)reflex recruitment curves during seated rest and seated active (10% maximal volitional contraction, MVC) conditions. Subthreshold transcranial magnetic stimulation (TMS) to the soleus M1 hotspot were paired with PNS (intensity 50% Hmax) using inter-stimulus intervals (ISIs) ranging from -10ms to 40ms, and the resulting H-reflex responses were used to calculate %modulation (conditioned H-reflex amplitude / unconditioned H-reflex amplitude * 100). For both groups, we evaluated the change in %modulation (Δ %modulation) from rest to active conditions at 3 ISIs (-1.5ms, 10ms, 40ms). Our preliminary results revealed larger activity-dependent Δ %modulation of spinal reflexes at -1.5ms in NOA (24.39% ± 8.67%) compared to stroke survivors (-3.79% ± 10.67%). This observation suggests that NOA show increased activity-dependent recruitment of direct, fast-conducting pathways, which is disrupted following stroke. In contrast, larger and positive activity-dependent Δ %modulation of spinal reflexes at 10ms was seen in post-stroke (10.68% ± 21.15%) versus NOA (-31.99% ± 42.01%), suggesting that stroke survivors may recruit more indirect, compensatory pathways during voluntary activation compared to NOA. Finally, at 40ms, an activity-dependent decrease in Δ %modulation was observed in both NOA (-32.10% ± 14.52%) and stroke (-20.98% ± 13.51%), which may be mediated by indirect, slower descending corticofugal and intraspinal inhibitory influences on spinal circuits. Our results suggest that activity-dependent modulation of descending circuits may be differentially affected by stroke, where neurotypical older adults show higher modulation of direct, faster conducting pathways compared to chronic stroke survivors, while chronic stroke survivors show higher modulation of indirect, slower pathways, likely mediated by brain stem and intraspinal circuits. Interestingly, at the longest ISI (40ms), both groups demonstrated an activity-dependent inhibition of spinal reflexes, which merits further investigation. Future studies will compare the current results with those in young adults, and associations between spinal reflex modulation with balance and gait function post-stroke.

A machine learning approach to predict stepping activity levels in individuals with chronic stroke

Emily Russell, Allison Miller, Darcy Reisman, Hyosub E. Kim, Vu Dinh

University of Delaware, Newark, Delaware, USA

Stroke results in numerous sequelae, including a reduction in walking ability and aerobic deconditioning.^1,2 This is problematic because reduced walking ability and aerobic deconditioning are associated with deficits in physical function,^3,4 depression,^5,6 and reduced self-efficacy.⁷ Consequently, many individuals with stroke are inactive⁸ and do not meet physical activity recommendations to maximize health benefits.^9,10 Two areas of relevance for the rehabilitation community are determining predictors of daily stepping activity that may inform whether an individual will meet aerobic activity thresholds^10,11 and whether an individual will be primarily a home or community ambulator.^3,12,13 However, this problem is complicated by the fact that many variables appear to influence activity levels after stroke,¹⁴ requiring advanced analytical techniques with the ability to handle large, heterogeneous datasets to determine the most relevant predictors. Thus, the purpose of this study was to use machine learning to determine which factors are consistently most important in predicting both home versus community ambulation (2500-step threshold)³ and whether physical activity recommendations are being met (5500-step threshold).¹⁰ Towards this end, we applied three different classes of machine learning algorithms (support vector machine, logistic regression, random forest) to a large dataset of 25 variables collected from 268 individuals with chronic stroke. We employed a two-step process to answer our research question. In step one, we used dimensionality reduction by means of lasso regularization to achieve a relevant subset of predictors. In step two, we computed a measure of feature importance for the remaining predictors for each algorithm and examined which predictors were deemed important across all three algorithms. Based on previous studies, we hypothesized that measures of physical capacity,^3,13,14 self-efficacy,¹⁵ and environmental factors^16,17 would be important predictors for both stepping thresholds. However, through our application of the three different machine learning algorithms to the data, we found that for the aerobic threshold, speed modulation was the only predictor that was consistently selected as important across all three algorithms. For the home versus community threshold, all three algorithms selected the 6-Minute Walk Test as the most important predictor. After regularization, the median prediction accuracies of each algorithm for the home versus community threshold were: logistic regression 64.9%, support vector machine 63.2%, random forest 68.1%. The prediction accuracies for each algorithm after regularization for the aerobic threshold were: logistic regression 73.5%, support vector machine 71.6%, random forest 74.1%. These results indicate that our machine learning approach offers a promising means of disentangling the complex relationship between laboratory-based measures of function and real-world activity identified in previous reports.^18,19 This work suggests that in order to improve a stroke survivor’s ability to walk within the community and meet physical activity guidelines, clinicians should target walking endurance and the ability to modulate walking speed, respectively.

Comprehensive Cardiac Rehabilitation Feasibility after Stroke (CCR FAST)

Amanda Herrmann ^1,2 , Ella Chrenka ^1,2 , Lauren O'Keefe ^1,2 , Bethany Bohnert ³ , Chad House ³ , William Nelson ³ , Leah Hanson ^1,2 , Haitham Hussein ^1,2,4

¹HealthPartners Neuroscience Center, St Paul, USA. ²HealthPartners Institute, Minneapolis, USA. ³Regions Hospital Cardiopulmonary Rehabilitation, St Paul, USA. ⁴Regions Hospital Comprehensive Stroke Center, St Paul, USA

Background: Few studies have examined the use of Comprehensive Cardiac Rehabilitation (CCR) or its impact on clinical endpoints in stroke patients. While CCR is not currently reimbursable for a stroke diagnosis, the overall goal of this project was to demonstrate the feasibility of stroke patients participating in a CCR program.

Methods: Patients ≥ 18 years who suffered a non-disabling ischemic stroke were recruited into this prospective feasibility study. Those who consented were enrolled into the CCR program, which consisted of 36 visits over 12 weeks. Each CCR visit included aerobic exercise, strength training, and patient education (lifestyle modification, nutrition, tobacco cessation and medication adherence), similar to the rehabilitation program for cardiac and pulmonary diseases. The primary endpoint was successful completion of the study protocol, which consisted of attending 75% of the CCR visits, completing a brain MRI and laboratory tests at the end of the study, and attending the end of study visit within nine months from enrollment. Patients were also sent an electronic CCR satisfaction survey.

Results: Fourteen patients were enrolled in this study, however, only 10 patients started CCR and only four patients completed the primary endpoint. The COVID-19 pandemic led to the temporary shutdown of unnecessary medical interventions and research operations. This prevented most patients from completing study procedures in the pre-specified timeframe. Of the 10 patients who began CCR, eight achieved 75% attendance for CCR visits, the end of study MRI, and attended the last visit. However, only four patients completed the end of study laboratory tests. Thus, only four patients met all the criteria of our primary endpoint. Of the four patients, all were male, the mean age was 65 ± 21, and three were white. When assessing clinical outcomes of the eight patients who completed CCR visits, none had a recurrent stroke, myocardial infarction or passed away. One patient had a silent stroke on the end of study MRI. Six of the eight patients who completed CCR responded to the satisfaction survey. All indicated they would recommended the CCR program to other stroke patients and did not have any safety concerns while participating.

Conclusion: While only four patients completed the entire study and eight patients completed CCR, we acquired numerous learnings from this study to apply to a larger future trial. The COVID-19 pandemic negatively impacted this study, by limiting patients’ ability to accomplish all criteria of the primary endpoint. Other logistical barriers with patients, such as other medical issues, transportation issues, and the time commitment were observed. Importantly, however, patients who participated in CCR were very satisfied with the care they received in the program, would recommend it to others, and felt they were provided with tools they needed to make lifestyle changes.

Post-Stroke Upper Extremity Motor Recovery Predicted by Sensorimotor Connectivity using EEG

Amanda Vatinno ¹ , Christian Schranz ¹ , Viswanathan Ramakrishnan ¹ , Leonardo Bonilha ¹ , Na Jin Seo ^1,2

¹Medical University of South Carolina, Charleston, USA. ²Ralph H. Johnson VA Medical Center, Charleston, USA

Background: Uncertain prognosis presents a challenge for therapists in determining the most efficient course of treatment for individual patients. Sensorimotor network connectivity may have prognostic utility because the integrity of the communication between sensorimotor cortices forms the basis for neuroplasticity and motor recovery. We hypothesized that patients with greater ipsilesional connectivity will have greater motor recovery with therapy. In addition, we hypothesized that use of sensory stimulation intended to increase neural communication (i.e. TheraBracelet) may assist recovery for patients with reduced ipsilesional connectivity.

Objective: The objective was to investigate if EEG connectivity predicts upper extremity motor recovery.

Methods: Retrospective analysis was performed for data from a pilot randomized controlled trial (n=12 stroke survivors). All participants underwent 2-week task-practice therapy, while receiving TheraBracelet stimulation for the treatment group and no stimulation for the control group. Recovery was quantified as change in Box and Block Test (ΔBBT) from baseline to post-therapy. Ipsilesional connectivity was obtained, in addition to contralesional/interhemispheric for comparison, using EEG pre-intervention. In particular, alpha connectivity was obtained for its involvement in sensory processing, attention allocation, and motor planning/execution. Conventional predictors (age, time post-stroke, lesion volume) were also investigated. The association between EEG and ΔBBT was examined using regression, controlling for group.

Results: The association between ipsilesional connectivity and ΔBBT differed by group (p=0.046). In the control group, greater ipsilesional connectivity was associated with greater recovery (r=0.44). In the treatment group, patients with greater ipsilesional connectivity had the same extent of improvement as the control group, while patients with lower initial connectivity had greater recovery (r=-0.57). In contrast, lower contralesional/interhemispheric connectivity was associated with greater recovery (p=0.017). Conventional predictors were not significant (p>0.05).

Conclusion: Pre-intervention sensorimotor connectivity may predict individual patients’ upper extremity motor recovery. In addition, patients with lower connectivity with less potential to recover may benefit from use of TheraBracelet that facilitates neural communication.

Explanation of recovery after stroke: the severely impaired are a distinct group

Anna Bonkhoff ¹ , Tom Hope ² , Danilo Bzdok ³ , Adrian Guggisberg ⁴ , Rachel Hawe ⁵ , Sean Dukelow ⁶ , François Chollet ⁷ , David Lin ¹ , Christian Grefkes ⁸ , Howard Bowman ⁹

¹Massachusetts General Hospital, Harvard Medical School, Boston, USA. ²University College London, London, United Kingdom. ³McGill University, Montreal, Canada. ⁴University of Geneva, Geneva, Switzerland. ⁵University of Minnesota, Minneapolis, USA. ⁶University of Calgary, Calgary, Canada. ⁷Centre Hospitalier Universitaire de Toulouse, Toulouse, France. ⁸University of Cologne, Cologne, Germany. ⁹University of Birmingham, Birmingham, United Kingdom

Introduction: Stroke can lead to different levels of motor impairments and the recovery thereof shows a great interindividual variety. A more comprehensive understanding of this recovery is an indispensable prerequisite to successfully guide clinical decision-making and neurorehabilitation treatments. However, a unified framework of the stroke recovery processes is challenged by the circumstance that the majority of recovery studies to date primarily considers patients with mild-to-moderate impairment levels, yet only very few studies address recovery patterns in those patients with severe initial impairment.¹ Additionally, recent work has highlighted substantial confounds underlying the concept of the proportional recovery rule, in particular due to mathematical coupling and ceiling effects.^2,3 After adjusting for these statistical confounds, only weak evidence for the originally assumed proportional recovery relationship in non-severely affected patients remained.⁴ Essentially, this finding questioned the validity of separately analyzing only this non-severe subgroup in the framework of the proportional recovery rule. We here aimed to develop a novel statistical framework to analyze recovery patterns in patients with severe and non-severe initial motor impairment and concurrently investigated whether recovery patterns differed between subgroups.

Methods: We employed Bayesian hierarchical regression to model upper limb Fugl-Meyer (FM) outcome scores three to six months after initial stroke. Analyses were focused on the explanation of recovery patterns. We here alleviated previous statistical confounds by only considering patients with FM-initial-scores below 45. Hierarchical levels were introduced to distinguish between severe/non-severe initial impairment, the FM-breakpoint between subgroups was systematically explored for values between FM-initial=5 and FM-initial=30. In model comparisons, we evaluated whether impairment-level-specific recovery patterns indeed existed or whether recovery patterns were the same across all degrees of initial impairment. Furthermore, we contrasted natural and drug-intervention-enhanced recovery patterns.

Findings: Recovery data was assembled from eight patient cohorts (n=489). Model comparisons suggested that recovery was best modelled by integrating two subgroups (breakpoints: FM-initial≤10 and FM-initial>10, R-squared=43%). Both of these subgroups featured a similar constant recovery portion, but were characterized by opposing proportional recovery contributions: while severely impaired patients showed a greater degree of recovery the smaller their initial impairment, non-severely impaired patients showed a greater degree of recovery the larger their initial impairment. After treatment with fluoxetine, the intervention-enhanced recovery of severely affected patients was characterized by higher amounts of constant and initial-impairment-specific contributions.

Conclusions: This work highlights the benefit of concurrently modelling recovery of severely and non-severely impaired patients and thus motivates more frequent combined analyses of patient subgroups. This inclusive approach has the potential to decrease the excess of conducted research on only non-severely affected patients. Our findings furthermore provide evidence that the severe/non-severe subdivision in recovery modelling is not an artifact of previous confounds and that neurorehabilitation interventions should be impairment-adapted.

Estimating impairment from functional task performance

Avinash Parnandi ¹ , Anita Venkatesan ¹ , Natasha Pandit ¹ , Audre Wirtanen ¹ , Emily Fokas ¹ , Grace Kim ² , Dawn Nilsen ³ , Heidi Schambra ¹

¹NYU School of Medicine, New York, USA. ²NYU Steinhardt, New York, USA. ³Columbia University Medical Center, New York, USA

Introduction: Quantifying upper extremity (UE) motor impairment after stroke is impractical, limiting our ability to tailor rehabilitation training in real time. The current gold-standard measure of impairment, the Fugl-Meyer Assessment (FMA), is time-consuming and requires a trained assessor. The FMA furthermore does not assess functional motions in real-world contexts, which is exactly where we aim our rehabilitation interventions. Here, we took initial steps to develop an approach to automatically quantify UE motor impairment during functional task performance.

Methods: We studied 51 chronic stroke patients (28F:23M; 57.7 (21.3-84.3) years old; 28L:23R paretic; FMA 43.1 (8-65)). We recorded upper body motion with 9 inertial measurement units (IMUs) while patients performed the FMA and a functional task (moving an object on a horizontal 8-target array). We trained a long short-term memory (LSTM) deep learning model to estimate FMA scores from the recorded motion (training set n=40; test set n=11; 4 LSTM layers with between-layer batch normalization; IMU data windows of 4s with slide of 1s). LSTM-generated impairment scores were computed from FMA motions or from functional motions. To ascertain the accuracy of the approach, we calculated the root mean square error (RMSE) and the Spearman correlation coefficient (ρ) between the LSTM scores and the FMA scores from a trained expert. We also examined whether the performance of particular classes of functional primitives (i.e. reach, transport, or reposition) would be sufficient to accurately estimate impairment.

Results: Using motions from the FMA performance, our approach estimated FMA scores within 1.1 points of a trained assessor. Using motions from the functional task performance, our approach estimated FMA scores within 1.6 points. Correlation values between the FMA scores and LSTM scores were ρ = 0.98 for FMA motions and ρ = 0.96 for functional motions. Among the three functional primitives, reaches were the most informative for estimating the impairment scores (RMSE: 1.9 points), followed by transports (RMSE: 2.1 points), and repositions (RMSE: 2.8 points).

Discussion: We present a new approach that uses sensor-based motion capture and deep learning to automatically estimate UE motor impairment. This approach has high accuracy and shows high concurrent validity with the FMA, even when it assesses unrelated functional motions. Thus, it may be possible to directly measure impairment from performance of real-world functional tasks, which the FMA does not offer. Estimating impairment during stroke rehabilitation would enable clinicians to tailor treatment strategy in real time.

NMDA receptor-mediated post-hypoxic potentiation in mice lacking glutamate antiporter, system x_c-

Bradley Stavros Heit, Bradley Stavros Heit, Alex Chu, Abhay Sane, Janet Richmond, David Featherstone, Alyssa McRay, John Larson

University of Illinois at Chicago, Chicago, USA

Ischemic stroke remains the leading cause of adult disability in the world. Efforts to reduce stroke severity, however, have been plagued by translational failure due to gaps in our understanding of cellular mechanisms leading to brain damage after metabolic insult. Loss of blood supply to brain tissue depletes neurons of energy, which induces a sequela of events within the neuronal network. Importantly, much of the “ischemic cascade” can be reproduced by transient deprivation of O₂ to in vitro hippocampal slices.

The glial-bound cystine/glutamate antiporter, system x_c-, with specific subunit xCT, supplies 60-80% of ambient extracellular glutamate in the brain . Using slice electrophysiology, we have previously shown that xCT-mediated glutamate release influences both the rapidity and synchrony of depolarizing events after anoxia. Both genetic deletion and pharmacological antagonism of system x_c- provided ischemic neuroprotection by increasing latency to anoxic depolarization (AD), attenuating AD wave amplitudes, and extending AD wave durations. The sudden onset and rapid regenerative nature of AD, however, may obscure subtle, but important, antecedent differences. Partial (“graded”) hypoxia, or reduced regional blood flow in vivo, is an ischemic condition that alters synaptic signaling, but is not noxious enough to induce glutamate excitotoxicity.

Graded hypoxia elicits a suppression of synaptic transmission proportional to the degree of oxygen deprivation, ATP breakdown, and accumulation of extracellular adenosine. The present study employed graded hypoxia to better understand ischemia-induced alterations in neuronal responsiveness from WT and xCT KO (xCT^-/-)mice. Hippocampal slices from both genotypes were prepared, incubated, subjected to 30 minutes hypoxia, and fully re-oxygenated. Although WT and xCT^-/- slices did not differ in hypoxia-induced synaptic suppression, mutant slices exhibited accelerated rate of recovery and post-hypoxic potentiation, which resembled LTP. Experiments using CPX, an A₁ adenosine antagonist, showed that this differential effect was not due to enhanced adenosine release in the xCT^-/-. Contrarily, further experimentation revealed that post-hypoxic potentiation in xCT^-/- mice is driven by activation of NMDARs and enhanced calcium influx. Thus, post-hypoxic changes in mutant mice are propagated by mechanisms similar to those of LTP. This latter finding was remarkable as xCT^-/- mice showed no differences in electrically-induced LTP compared to WT. Taken together, these data confirm system x_c- as a salient regulator of neuronal responses during ischemia and a therapeutic target ripe for exploration.

The role of glutamate antiporter, system xc-, in the ischemic cascade

Bradley Stavros Heit, Bradley Stavros Heit, Alex Chu, Abhay Sane, Janet Richmond, David Featherstone, John Larson

University of Illinois at Chicago, Chicago, USA

Ischemic stroke remains the third leading cause of death and leading cause of adult disability worldwide. Efforts to reduce stroke severity, however, have been plagued by translational failure due to gaps in our understanding of cellular mechanisms leading to brain damage after metabolic insult. Loss of blood supply to brain tissue (ischemia) depletes neurons of energy (ATP), which induces an excitatory effect in the synaptic network marked by the excessive release of glutamate. Within minutes, this excitation results in irreversible tissue damage triggered by anoxic depolarization (AD) - an electrophysiological event with predictive value for stroke outcome. Importantly, much of the “ischemic cascade” can be reproduced by transient deprivation of O₂ to in vitro hippocampal slices.

The cystine/glutamate antiporter, system x_c-, with specific subunit xCT, is the primary source of ambient extracellular glutamate in the brain. This glial-bound transport system serves as a source of cystine, which is intracellularly converted to cysteine - the rate-limiting substrate for glutathione synthesis. The high rate of O₂consumption in the brain renders this antiporter vital to antioxidant defense, and its expression is rapidly upregulated during oxidative stress. However, the obligate release of glutamate into the extracellular space, accompanying the uptake of cystine, could exacerbate the synaptic response during O₂ deprivation. We therefore employed hippocampal slice electrophysiology to investigate the role of system x_c- during total oxygen deprivation (anoxia). Using a battery of electrophysiological assays, we compared the anoxia tolerance of transgenic mice lacking a functional system x_c- (xCT^-/- mice) to WT controls. Our results show that both genetic deletion and pharmacological antagonism of system x_c- increase latency to AD and attenuate depolarizing waves. Moreover, xCT^-/- mice exhibit no alteration in paired-pulse facilitation, thus confirming their anoxia tolerance is not a function of presynaptic calcium release. These data suggest that anoxia-induced excitation in the synaptic network is enhanced due to the antiporter’s putative effect on extracellular glutamate levels. Experiments where extracellular glutamate concentrations were manipulated, as well as experiments performed in conditions of glutamate receptor antagonism, further confirm this hypothesis. Taken together, our findings reveal that xCT-mediated glutamate release is a salient driver of the ischemic cascade.

The deleterious effects of stroke pathogenesis remain, to a large degree, intractable and irreparable. Despite the neuroprotective effects of glutamate receptor antagonism for in vivo and in vitro ischemia models, clinical trials with pharmacological antagonists have shown poor translational value. Perhaps future efforts should focus on antagonizing the source(s) of glutamate as opposed to the target of its action. Our report contributes to the growing breadth of research implicating the cystine/glutamate antiporter as an attractive target for ameliorating or preventing the devastating effects of cerebral ischemia.

Ipsilateral and paretic motor evoked potentials have delayed onset latency in individuals with chronic stroke

Brice Cleland, Emily Sisel, Sangeetha Madhavan

University of Illinois at Chicago, Chicago, USA

Introduction: Reduced impairment after stroke may occur through recovery of damaged pathways or compensation by other neural pathways, including ipsilateral pathways from the contralesional hemisphere. Studies characterizing the onset latency and duration of ipsilateral motor evoked potentials (MEPs) evoked by transcranial magnetic stimulation (TMS) can provide insight into the role of ipsilateral pathways, but work in the paretic lower limb has been minimal. This study assessed MEP onset latency and duration in the lower limb after stroke and compared ipsilateral vs. contralateral and paretic vs. non-paretic MEPs. We hypothesized that: 1) onset latency and duration would be longer for ipsilateral than contralateral MEPs, and 2) onset latency would be longer for the paretic than the non-paretic limb.

Methods: Data were collected during a pre-test for a randomized controlled trial. During isometric contractions of the tibialis anterior, TMS was applied at 120% of active motor threshold to the ipsilateral and contralateral hemisphere. MEP onset/offset latency were characterized as when electromyography (EMG) amplitude went below/above 25% of background for >5 ms. MEP duration was the time from MEP onset to offset. We performed repeated measures analysis of variance (ANOVA) with within subject factors of limb (paretic and non-paretic) and stimulation side (ipsilateral and contralateral).

Results: 35 participants with chronic stroke who had ipsilateral and contralateral MEPs in both limbs were included in the analysis. MEP onset latency was longer in the paretic than the non-paretic limb (32.3 (5.6) ms vs. 26.3 (3.4) ms; mean difference=6.0 ms, 95% confidence interval (CI): 4.7,7.2; p<0.001) and was longer after ipsilateral than contralateral stimulation (30.2 (5.7) ms vs 28.4 (5.1) ms; mean difference=1.8 ms; 95% CI: 1.1,2.6;p<0.001). Duration was longer in the paretic than the non-paretic limb (54.7 (12.0) ms vs. 45.5 (8.0) ms; mean difference=9.2 ms, 95% CI: 5.8,12.4; p<0.001) and was longer after contralateral than ipsilateral stimulation (52.7 (10.9) ms vs. 47.5 (10.8) ms; mean difference=5.2 ms, 95% CI: 3.0,7.3; p<0.001).

Discussion: Delayed MEP onset latency and duration in the paretic lower limb may reflect slower conduction time or peripheral changes leading to smaller MEPs that are difficult to distinguish from background activity and excite. Alternatively, longer paretic MEP onset latency and duration may reflect the activation of pathways containing a greater number of synapses, such as spinal interneuronal pathways and corticosubcorticospinal pathways. Delayed MEP onset latency and shorter duration for ipsilateral vs. contralateral MEPs may reflect the activation of the uncrossed corticospinal tract, which requires a greater stimulus to activate short latency D-waves and later I-waves. This finding also may reflect the activation of corticosubcorticospinal pathways, which have additional synapses. Overall, results provide insight into the potential contribution of ipsilateral pathways to stroke recovery, which has important implications for neurorehabilitation.

LUXURY PERFUSION SCORE FOR PREDICTING CLINICAL OUTCOME IN ANTERIOR CIRCULATION ISCHEMIC STROKE: A PILOT STUDY

Camilla Russo ¹ , Flavio Giordano ¹ , Giuseppe Leone ¹ , Massimo Muto ¹ , Gianluigi Guarnieri ¹ , Eduardo Gragnano ² , Laura Lombardi ² , Donatella Franco ² , Gennaro Ambrosanio ¹ , Mario Muto ¹

¹Department of Neuroradiology, A.O.R.N. Cardarelli, Naples, Italy. ²Department of Precision Medicine, University of Campania “Luigi Vanvitelli”, Naples, Italy

Purpose: Recanalization in acute ischemic stroke with flow restoration prevents from brain tissue infarction and haemorrhagic transformation. Among recanalization techniques, mechanical thrombectomy (MT) is the golden standard in case of large artery occlusion, especially for anterior circulation. After recanalization, sudden increase in blood flow and impaired auto-regulation may result in a paradoxical effect known as luxury perfusion (LP). LP refers to abnormal vessel dilation in the ischemic territory occurring within few hours from stroke onset, generally expected after recanalization; paradoxical hyperperfusion is due to aberrant vessel recruitment and results in non-nutritive flow. Blood–brain–barrier damage contributes to this contrast leakage, leading to increased density of infarcted area on CT scan performed after recanalization. However, the relationship between LP, infarction and clinical outcome still remains largely unclear. Here we investigated the correlation of long-term clinical outcome with different CT parameters in a group of patients undergone MT for anterior circulation ischemic stroke.

Methods: In this mono-centred retrospective study, we included 89 patients with anterior circulation ischemic stroke (49M:40F;mean age:71.9±10.4y) selected for MT. All patients underwent: non-contrast computed tomography (NCCT) for ASPECTS calculation; multi-phase CT angiography (mCTA) for occlusion identification and collaterals scoring; automated CT perfusion (CTP) with RAPID software to quantify core infarct and penumbra; angiography for MT. After thrombectomy, post-procedural CT was performed to assess LP and exclude complications; a further NCCT control was performed 24h after thrombectomy. All patients received neurological examination with NIHSS at baseline, and 3-months outcome evaluation by telephonic assessment of modified Ranking Scale (mRS). Post-procedural CTs were revised by three neuroradiologist in consensus to assess LP score, a 10-point topographic CT scale borrowed from ASPECTS and used to assess increased density areas in the involved vascular territory after recanalization. Spearman rank correlation was used to measure the association of mRS with clinical and imaging findings at baseline/after recanalization, in order to identify a reliable long-term outcome predictor; statistical analysis was performed using Xlstat (2019.7).

Results: No correlation emerged between mRS and baseline/24h ASPECTS, mCTA scale for collaterals assessment (p>0.1), and baseline NIHSS (p>0.05). When testing the relation between mRS and CTP, a significant correlation was observed for core infarct and ischemic penumbra volumes (p<0.05), while no association was found with hypoperfusion index (p>0.1). Finally, a significant correlation emerged between post-revascularization LP score and mRS (p<0.05).

Conclusions: After recanalization, LP score could represent an early and widely applicable schematic approach for predicting long-term clinical outcome on post-procedural CT in patients who underwent MT for anterior circulation ischemic stroke, especially when other factors such as CTP parameters are not available. However, predictive values and further validation on larger samples are imperatively required to test LP score reliability and reproducibility in daily clinical practice.

Remedial training of the less-impaired arm in chronic stroke survivors with severe paresis improves functional independence: A pilot study

Candice Maenza ^1,2 , David A. Wagstaff ² , Carolee Winstein ³ , David C. Good ¹ , Robert L. Sainburg ^2,1

¹Penn State College of Medicine, Hershey, USA. ²Pennsylvania State University, University Park, USA. ³University of Southern California, Los Angeles, USA

The ipsilesional arm of stroke patients often has functionally limiting deficits in motor control and dexterity that depend on the side of the brain that is lesioned and that increase with the severity of paretic arm impairment. However, remediation of the ipsilesional arm has yet to be integrated into the usual standard of care for upper limb rehabilitation in stroke, largely due to a lack of translational research that examines the feasibility and potential efficacy of ipsilesional-arm intervention. We now ask whether intense ipsilesional-arm training, tailored to the hemisphere-specific nature of ipsilesional-arm motor deficits in participants with severe contralesional paresis, improves ipsilesional arm performance and generalizes to improve functional independence. We assessed the effects of this intervention on ipsilesional arm unilateral performance (Jebsen-Taylor Hand Function Test), ipsilesional grip strength, contralesional arm impairment level (Fugl-Meyer Assessment), and functional independence (Functional Independence Measure) (N = 13). Intervention occurred over a three-week period for 1.5 hours each week. All sessions included virtual reality tasks that targeted the specific motor control deficits associated with either left or right hemisphere damage, followed by intense and graded dexterity training in real-world tasks. We also exposed participants to 3 weeks of sham training to control for the non-specific effects of therapist visits and interactions. We conducted 5 test-sessions that consisted of 2 pre-tests and 3 post-tests: 1) directly followed the 3-week intervention, 2) directly followed the 3-week sham training, 3) the final posttest was 3 weeks later, to examine short term durability of training. Our results indicate substantial improvements in non-paretic arm performance, without detriment to the paretic arm, that transferred to improve functional independence in all three posttests, indicating durability of training effects for at least 6 weeks. We provide evidence for establishing the basis for a rehabilitation approach that focuses on remediation of each arm individually. This study was originally a crossover design; however, we were unable to complete the second arm of the study due to the COVID-19 pandemic. We report the results from the first arm of the planned design as a longitudinal study.

Upper limb performance plateaus before or with impairment and capacity post stroke

Catherine E. Lang, Kimberly J. Waddell, Jessica Barth, Carey L. Holleran, Michael J Strube, Marghuretta D. Bland

Washington University School of Medicine, Saint Louis, USA

Wearable movement sensors now allow for direct measurement of upper limb performance in daily life after stroke outside of the clinic. Emerging data suggest that upper limb capacity measures taken in the clinic or lab are not a good proxy for upper limb performance in daily life, and open up new questions about trajectories of upper limb performance post stroke. The purpose of the current study was to map the trajectory of upper limb performance and its relationships to impairment, capacity, and other factors over the course of stroke recovery.

Sixty-two participants with first-ever stroke and residual upper limb paresis were assessed at 2, 4, 6, 8, 12, 16, 20, and 24 weeks after injury. The main assessments captured upper limb impairment (Fugl-Meyer), capacity for activity (Action Research Arm Test), and performance of activity in daily life (accelerometer variables of use ratio and hours of paretic limb activity). Additional data on demographics, cognition, hemispatial neglect, depressive symptomatology, and rehabilitation service utilization were collected. Using hierarchical longitudinal analyses, we modeled individual trajectories over time for each main assessment and tested the moderating effects of various factors on these.

Individual trajectories were best fit with a 3-parameter logistic model, capturing the rapid growth earlier after stroke within the longer data collection period. As expected, plateaus (defined as 90% of asymptote) in impairment (bootstrap mean ± SE: 32 ± 4 days post stroke) preceded those of capacity (41 ± 4 days) by about 1 week. Surprisingly, plateau in performance as measured by the use ratio (24 ± 5 days) preceded plateaus in impairment and capacity. Plateau in performance, as measured by hours of paretic activity (41 ± 6), was timed similarly to the plateau in capacity and insignificantly lagged impairment.

Better capacity lead to quicker and better performance recovery. Eventual performance achieved was better but took longer for those with a concordant dominant and paretic limb. More depressive symptoms led to lower eventual upper limb performance but did not affect rate. More early cognitive deficits led to later recovery of performance measured by paretic hours, but did not reach significance for the use ratio. Other factors evaluated did not moderate upper limb performance trajectories.

Upper limb performance in daily life stabilized between 3 and 6 weeks post stroke on average. Plateaus in performance did not lag plateaus in impairment and capacity. Thus, individuals with stroke achieve a stable pattern of upper limb use in daily life surprisingly early and often before neurological and functional recovery stabilize. Given that a key purpose for referring to and participating in upper limb rehabilitation services is to improve performance in daily life, knowledge gained from this study informs the timing and content of rehabilitation services.

Neural re-organization after upper extremity rehabilitation therapy with sensory stimulation in chronic stroke survivors

Christian Schranz, Amanda Vatinno, Viswanathan Ramakrishnan, Na Jin Seo

MUSC, Charleston, USA

Background: Stroke is a leading cause of disability worldwide [1]. Many stroke survivors suffer from upper-extremity motor impairment [2]. Facilitating neuroplasticity of the sensorimotor network in the brain could improve post-stroke recovery [3]. Previous research employed a novel device called TheraBracelet, that uses subsensory stimulation to facilitate neural communication and promote functional recovery [4]. While this intervention has been shown a promise to improve motor recovery, little is known about its neural mechanism [4]. This study aims to investigate the effect of using this stimulation during therapy on cortical sensorimotor activity to improve our understanding of neural processes underlying motor recovery.

Methods: Twelve chronic stroke survivors underwent 2-weeks of upper extremity task-practice therapy, while wearing a vibrator on their paretic wrist. They were randomized to either a treatment group that received imperceptible vibratory stimulation, or control group with no stimulation during therapy. EEG during paretic hand grip and its preparation phase was measured at pre-intervention, post-intervention, and 3-week follow-up. Specifically, EEG connectivity and power for the sensorimotor network including premotor, primary motor and primary sensory cortices of both hemispheres were examined. Change was evaluated through multifactorial ANOVA.

Results: Significant group by time by phase interactions were observed for both connectivity and power (p<0.05). Posthoc analysis showed that the treatment group increased connectivity in the sensorimotor network bilaterally from pre- to post-intervention for both phases. Connectivity returned to baseline at follow-up. In addition, the treatment group reduced power modulation bilaterally at post-intervention, which was maintained at follow-up. On the contrary, the control group showed a slight increase of the connectivity in the preparation phase only at follow-up compared to preintervention and no power change.

Conclusions: These results show the way sensory stimulation influences neuroplasticity. The increased connectivity post intervention in the treatment group may indicate neuroplasticity specific to motor learning [5]. Return of connectivity to the baseline level at follow-up may be due to no further motor learning occurring during the follow-up period. In addition, reduced power modulation in the treatment group may indicate lessened effort for grip with the paretic hand with learned motor skills at post-intervention and follow-up [6]. These findings complement the clinical results reported in Seo et al. (2019) in which the treatment group significantly improved clinical upper-extremity motor function at post-intervention and follow-up while the control group did not [7]. These findings suggest EEG may serve as a biomarker for motor learning and neuroplasticity and may provide potential neural targets for novel therapy approaches.

A Gamified Electromyographic Computer Interface to Measure Individual Motor Control Impairments Across Multiple Time Points

Danielle Marouni ^1,2 , Yiyun Wang ¹ , Nathan Pinnette ¹ , Ania Busza ³

¹University of Rochester, Rochester, NY, USA. ², ³University of Rochester Medical Center, Rochester, NY, USA

Background: Upper extremity (UE) disability is common after stroke. Prior studies in individuals with hemiparesis due to stroke have identified four distinct impairments of motor control: (1) decreased maximal muscle activation, (2) delayed muscle activation, (3) motor fatigue, and (4) dis-coordinated muscle activation, including co-activation of antagonistic muscle groups. In order to develop impairment-oriented interventions for motor recovery, more information is needed about the emergence and prevalence of these impairments. We have developed an electromyographic (EMG) computer interface to collect EMG signals from subjects perform repetitive muscle activations, with goal of characterizing the evolution of each impairment over the first 3 months post-stroke.

Objective: To collect EMG data from healthy individuals using our EMG computer interface to (1) create a healthy control data set, and (2) measure the intra-subject variability in healthy controls across multiple time points, to assess whether the system can be used to compare relative motor impairments in a longitudinal study of individuals with recent stroke.

Design/Methods: Our EMG computer interface uses surface EMG signals from the wrist flexor and extensor muscle groups to control a simple computer game. Successful gameplay requires multiple isometric muscle contractions at precise time points, each lasting 3.5 seconds. EMG data are analyzed to identify maximum EMG values, muscle activation delay, co-activation of antagonist muscle groups, and motor fatigue. EMG data from health young (18-30 year old) and older (60-90 year old) adults is being collected to create a healthy control data set for future comparisons. A subgroup of individuals are recorded multiple times, separated by several days, to measure intra-subject variability. In parallel, the system is being used to collect EMG data from subjects with hemiparesis due to recent (<4 weeks) stroke.

Results: Our system is well-tolerated and collects relatively large amounts of data from healthy controls and individuals with recent stroke. EMG data collected from healthy controls on two recording sessions separated by several days shows low intra-subject variability. Preliminary data collected from individuals with hemiparesis due to recent (<4 weeks) of stroke show increased delay in activation in their affected arm as compared to their unaffected arm. Some individuals show increased co-activation and increased motor fatigue in their paretic arm, especially when the required activation threshold is increased.

Conclusions: Our EMG-controlled computer interface is well tolerated and can be used to collect information about relative motor impairment levels in an individual across multiple time points. Our preliminary results are consistent with prior studies’ reports that some individuals with hemiparesis due to stroke have high levels of specific motor impairments such as delay in muscle activation and abnormal co-activation. Further studies will assess the prevalence and individual trajectory of each motor impairment over the first 3 months post-stroke.

Arm Motor Recovery After Ischemic Stroke: A Focus on Clinically Distinct Trajectory Groups

Danielle Kline ¹ , David Lin ² , Alison Cloutier ² , Kelly Sloane ² , Kristin Parlman ² , Jessica Ranford ² , Matthew Picard-Fraser ¹ , Annie Fox ¹ , Leigh Hochberg ^2,3,4 , Teresa Kimberley ¹

¹MGH Institute of Health Professions, Boston, MA, USA. ²MGH, Boston, MA, USA. ³Brown University, Providence, RI, USA. ⁴VA Medical Center, Providence, RI, USA

Background and Purpose: Recovery of arm function post-stroke is highly variable with some people experiencing rapid recovery but many experiencing slower or limited functional improvement. Current stroke prediction models provide some guidance for clinicians regarding expected motor outcomes at a specific timepoint post-stroke. However, the models do not capture recovery rates which affects the utility of the models in discharge planning. This study developed a novel approach to defining recovery groups based on arm motor recovery trajectories post-stroke. Additionally, between-group differences in baseline characteristics and total therapy hours were explored.

Methods: A retrospective cohort analysis was conducted on 40 participants with arm weakness who were assessed 1-week, 6-weeks, 3-months, and 6-months after an ischemic stroke. Arm recovery trajectory groups were defined based on the timing of a person’s 10-point minimal clinically important difference (MCID) change in the Fugl-Meyer Assessment Upper Extremity (FMA-UE) across the first 6-months post-stroke. The MCID represents the lowest amount of change on the FMA-UE that is meaningful to patient care management. Three recovery trajectory groups were defined (Fast Recovery (n=19), Extended Recovery (n=12), Limited Recovery (n=9)). The Fast Recovery group demonstrated improvement of at least the FMA-UE MCID point change (≥10) before the 6-week assessments and then plateaued with less than a 10-point improvement thereafter. The Extended Recovery group gained at least an FMA-UE MCID point improvement (≥10) after completion of the traditional 6-week rehabilitation period and achieved any amount of FMA-UE change between the 1-week and 6-week assessments. The Limited Recovery group failed to achieve the FMA-UE MCID (<10) at all follow-up assessment time points. Between-group differences in baseline characteristics and total therapy hours (physical therapy and occupational therapy) were assessed. Associations between baseline characteristics and group membership were also determined.

Results: Three baseline characteristics were associated with trajectory group membership: FMA-UE, NIH Stroke Scale, and Barthel Index. The Fast Recovery group received the least total therapy hours 6-weeks to 6-months. No differences in total therapy hours were observed between Extended and Limited Recovery groups at any time points.

Discussion and Conclusions: Three clinically relevant arm recovery trajectory groups were defined based on the timing of FMA-UE MCID changes between 1-week and 6 months post-stroke. Baseline arm impairment, overall stroke severity, and ADL dependence were associated with group membership. Our novel method of assessing the trajectory of recovery from acute to chronic stroke could move the field toward better prediction models by recognizing that the path to recovery is not on the same timeline for everyone. Moreover, a better understanding of the characteristics that are associated with arm recovery trajectories across the continuum of care could improve the individualization of patient discharge plans and the optimization of limited therapeutic resources post-stroke.

Too much to handle: performance of dual-object primitives is limited in the nondominant and paretic upper extremity

Emily Fokas, Avinash Parnandi, Anita Venkatesan, Natasha Pandit, Audre Wirtanen, Heidi Schambra

NYU School of Medicine, New York, USA

Introduction: Activities of daily living (ADLs) are performed through a sequence of fundamental units of motion, called primitives. We previously observed that during ADLs, one upper extremity (UE) may engage two objects simultaneously, such as turning on a faucet while holding a toothbrush. These dual-object primitives (DOPs) may demand increased neural resources, as they likely entail the simultaneous execution of two motor plans. Skilled movement by the nondominant healthy UE or the paretic UE has also been found to require increased neural activity. We posited that performance of DOPs would exceed the neural resources available to the nondominant or paretic side, reducing their performance on these sides. We also predicted that the frequency of DOP performance by the paretic UE would relate to its degree of motor impairment.

Methods: We studied 19 right-hand dominant healthy subjects (10M:9F; 62.0 ± 13.6 years) and 43 premorbidly right-hand dominant stroke subjects (23M:20F; 24L:19R paretic; 57.5 ± 14.5 years; 5.7 ± 6.5 years post stroke). We evaluated subjects on the UE Fugl-Meyer Assessment (FMA) and videotaped their performance of a feeding and toothbrushing task. We analyzed the videos to extract the incidence and count of DOP performance by each UE. To control for dominance and paresis, we normalized DOP counts to the total number of primitives performed by the UE. We used two-tailed Fisher’s Exact tests to compare the incidence of DOPs performed by each UE, and Spearman’s correlation to examine the relationship between FMA score and DOP frequency.

Results: In healthy subjects, the incidence of DOPs was lower on the nondominant than dominant side (12/19 vs. 19/19; p<0.01). In stroke subjects, the incidence of DOPs was lower on the paretic than nonparetic side (19/43 vs. 43/43; p<0.01). The laterality of paresis did not affect whether that UE would perform DOPs (11/19 dominant paretic vs. 8/24 nondominant paretic; p=0.132). In stroke subjects, lower FMA scores were related to a lower frequency of DOP performance on their paretic UE (ρ=0.368, p=0.015).

Discussion: Our results suggest that UE laterality and impairment may impact DOP performance in healthy and stroke subjects, respectively. DOPs were less commonly performed by the nondominant UE and the paretic UE, and worse impairment was associated with lower DOP performance. We speculate that engaging two objects simultaneously requires additional neural resources that are unavailable to the nondominant or injured motor network. It is conceivable that the return of DOP performance by the paretic UE may track with the availability of a recovered neural substrate.

Impact of Amantadine Hydrochloride on Hospitalized Acute Ischemic and Hemorrhagic Stroke Patients

Enzo Plaitano ^1,2 , Rebecca Scharf ³ , Pakinam Aboutaleb ¹ , Emma Jost-Price ⁴ , Deborah Green-LaRoche ⁴

¹Tufts University School of Medicine, Department of Neurology, Boston, USA. ²Boston University, Undergraduate Program in Neuroscience, Boston, USA. ³Tufts University School of Medicine, Boston, USA. ⁴Tufts University School of Medicine, Departments of Neurology and Neurosurgery, Boston, USA

Introduction: Stroke is the fourth leading cause of disability in the United States (1). Patients are often bedridden during recovery and thus cannot participate in inpatient rehabilitation, which can lead to worsened long-term outcomes (2). In response, amantadine hydrochloride has been prescribed as a neurostimulant for stroke patients and was theorized to promote wakefulness during inpatient recovery (3). Historically, amantadine has been used to relieve symptoms of Parkinson’s Disease, including dyskinesia (4). This medication was then prescribed to traumatic brain injury patients in the acute rehabilitation setting to decrease lethargy and promote participation in therapy (5). The specific impact of amantadine in hospitalized patients who have had ischemic or hemorrhagic strokes still remains unknown. This study aims to determine the association between amantadine and patients’ motor, communicative, cognitive, and behavioral abilities at the time of hospital discharge.

Methods: Retrospective review included patients with acute ischemic or hemorrhagic stroke who presented to a comprehensive stroke center between 2016 and 2019 and were prescribed amantadine hydrochloride for neurostimulation (i.e., wakefulness and language promotion (n=34)). Patients were grouped as responders–those who had a ≥3-point increase in Glasgow Coma Scale (GCS) or clinical improvement in wakefulness reported in provider or therapist notes within 9 days of the initial dose–versus nonresponders who did not meet these criteria (3). The relationship between amantadine response status and neurological impact was evaluated with multivariable logistic regression models, which controlled for potential confounders.

Results: Thirty-four stroke patients received amantadine with 19 (55.9%) amantadine responders (63.2% male, 63.0 mean age, 63.2% hemorrhagic) versus 15 (44.1%) nonresponders (53.3% male, 66.0 mean age, 66.7% hemorrhagic). 29 (85.3%) patients received amantadine for wakefulness, while 5 (14.7%) received the drug for both wakefulness and language promotion. Among responders, the average time to responder status was 2.82 ± 0.439 days and the range was 2-9 days. In multivariable models, amantadine responders were associated with an increase in success rate (%) following 1 or 2-step commands [b=42.325, 95%CI (10.60, 74.05), p=0.012], increase in GCS [b=5.306, 95%CI (2.876, 7.736), p=0.003], and decrease in modified Rankin Scale (mRS) [b=-0.4779, 95%CI (-0.9245, -0.03141), p=0.037] at discharge compared with nonresponders. Among all recipients, responder status was positively associated with active participation in therapy sessions [b=0.477, 95%CI (0.0465, 0.9077), p=0.032], and fluent language output [b=0.844, 95%CI (0.5884, 1.100), p=0.000002], with 79.0% of responders and 0.00% of nonresponders having fluent language at discharge.

Conclusions: Retrospective analyses demonstrated that response to amantadine was significantly associated with an increase in participation and following commands during therapy, improved language abilities, higher GCS, and lower mRS scores at hospital discharge. Amantadine hydrochloride may therefore prove beneficial to hospitalized stroke patients.

Feasibility of a Self-Directed Home Therapy Program for Stroke Survivors

Gabrielle Scronce, Amanda Vatinno, Corey Morrow, Allison Pennington, Na Jin Seo

Medical University of South Carolina, Charleston, USA

Rehabilitation of hand function post-stroke can require extensive amounts of therapy. Optimal outcomes are more likely to be achieved when individuals practice therapy independently at home in addition to regular therapy sessions.

The objective of this study was to determine feasibility of a self-directed home therapy component of a rehabilitation program for stroke survivors.

Data from 2 separate home interventions of upper extremity task-specific practice were evaluated for this study. A total of 13 participants with chronic stroke recorded their home performance of prescribed therapeutic activities 5 days per week for a period of 4-6 weeks. Additionally, 9 participants recorded the time that they wore an investigational TheraBracelet stimulation device daily for 6 weeks. An occupational therapist monitored therapy progress and met with participants to collaboratively select therapy tasks and help address barriers to therapy adherence. Adherence was calculated as the percentage of prescribed repetitions (200 or 300 repetitions of selected functional tasks) completed as well as the percentage of prescribed duration (8 hours) that the TheraBracelet was worn. Relationships between adherence and baseline characteristics were examined using Pearson correlations.

The home program was feasible, as average adherence was 80.7% for upper extremity task-specific practice repetitions and 116.9% for wearing TheraBracelet. Adherence varied by participant, with a moderate, positive correlation between upper extremity task-specific practice adherence and baseline arm use measured by the Motor Activity Log, indicating that participants with more functional use of their affected upper extremity prior to the study were more adherent to exercise recommendations. A moderate, negative correlation between adherence to upper extremity task-specific practice and adherence to Therabracelet wear may suggest that individuals who have difficulty adhering to exercise recommendations would benefit from alternative therapeutic interventions. Future direction includes determining mechanisms to increase adherence to rehabilitation recommendations among all participants and to investigate TheraBracelet for upper extremity recovery post-stroke.

Developing newly emerging intermuscular coordination patterns through an electromyographic signal-guided exercise in the upper extremity

Gang Seo ¹ , Jeong-Ho Park ² , Hyung-Soon Park ² , Jinsook Roh ¹

¹Department of Biomedical Engineering, University of Houston, Houston, USA. ²Department of Mechanical Engineering, Korea Advanced Institute of Science and Technology, Daejeon, Korea, Republic of Korea

Along with the classic stroke-induced motor deficits, such as paresis, muscle weakness, and abnormal muscle tone, impaired intermuscular coordination has recently been targeted in the field of motor rehabilitation after stroke. Using the concept of muscle synergies, characteristic co-activation patterns of a group of muscles, relatively recent works have characterized the abnormal intermuscular coordination in the upper extremity (UE) by applying dimensionality reduction methods. The muscle synergies affected by stroke involve the altered coupling of elbow and shoulder muscles as well as the abnormal coactivation of three heads of deltoid muscles under isometric conditions. However, it is still unclear whether targeting or normalizing the altered muscle synergy(ies) would resolve the stroke-induced motor impairment and improve functional movement. Therefore, we hypothesized that modifying the abnormal muscle synergy through rehabilitation exercise can improve impaired motor function of the UE after stroke. To validate the modifiability of muscle synergy, first, we developed an electromyographic (EMG) signal-guided training protocol that targeted to develop newly emerging synergies by altering an elbow flexor synergy observed in healthy subjects. Four research participants were trained to activate two major elbow flexor muscles of the dominant arm, biceps and brachioradialis, in isolation under the isometric condition for six weeks. At the week zero, two, four, and six of the training, the subject performed an isometric reaching in a three-dimensional force space to assess any potential changes in intermuscular coordination. While the subjects performed the motor assessment task, EMGs of eight major UE muscles and three-dimensional forces were collected simultaneously. A non-negative matrix factorization algorithm was applied to the EMGs to identify muscle synergies. The preliminary results showed that elbow flexor synergy could be altered after two-to-three weeks of the training. The targeted muscle pair, biceps and brachioradialis, which were activated synergistically before the training, activated in isolation from each other and formed new synergistic muscle groups with other muscles. As the new muscle synergies emerged, the accuracy and efficiency of the motor control of the trained arm were enhanced. Participants could still recruit habitual muscle coordination if they intended to do so. The newly emerging synergies could be activated if the participants intended to use the newly learned motor skill. Once the new intermuscular coordination patterns were developed through the isometric exercise, they remained throughout the rest of the training period. These results suggest that the proposed EMG-guided training under isometric conditions can increase the repertoire of readily available muscle synergies and improve motor control by inducing the activation of newly emerging muscle coordination patterns. We expect that the proposed exercise would improve the UE motor function of stroke survivors by expanding their repertoire of intermuscular coordination patterns, which will be tested as the next step.

Cross-validation of laboratory-based measures for quantifying flexion synergy in individuals with chronic stroke

Grace Bellinger, Michael Ellis

Northwestern University, Chicago, USA

Flexion synergy is a trademark movement pattern that emerges following stroke in humans. This abnormal muscle coactivation pattern is driven by volitional shoulder abduction and manifests as flexion across all distal joints. The Fugl-Meyer Motor Assessment is widely used in clinical research to measure generalized stroke impairment. Several laboratory-based measures have been used to quantify flexion synergy, but few have evidence supporting validity. The purpose of the study was to evaluate the criterion validity of four dynamic measures of flexion synergy within a cohort of thirty-seven individuals with chronic stroke using the Upper Extremity Fugl-Meyer Motor Assessment. Collectively, the cohort was 58.43±10.50 years of age, 11.89±8.81 years post-stroke, and had an average Fugl-Meyer score of 26. The four lab-based metrics included biomechanical and physiological measurements of the more affected upper limb in dynamic motor tasks. Each participant was coupled to a mechatronic device and completed a set of at least five center-out reaches while lifting against 50% of their maximum isometric shoulder abduction strength. Reaching distance was defined as maximum planar reaching distance normalized by target distance, which was dependent on arm length. Flexion synergy-related biceps EMG was also quantified during the reaching trials as the instantaneous activity following shoulder abduction but prior to reaching onset. The EMG value for each trial was normalized by maximum biceps EMG activity and the median was reported due to varying numbers of trials across participants. The final protocol identified two distinct force thresholds defining the boundaries of flexion synergy expression. The takeover and emergence thresholds were the heaviest shoulder abduction loads at which each participant could acquire standardized near and far targets, respectively. These values were normalized by maximum shoulder abduction strength. Fugl-Meyer score was significantly correlated with emergence threshold (r=0.464, p=0.015) and reaching distance (r=0.405, p=0.016). The moderate correlation coefficients are likely due to the fact that the Fugl-Meyer assesses movement both in and out of synergy while the lab-based measures focus on the ability to move out of flexion synergy. The correlation with takeover threshold was not statistically significant, and the measure demonstrated a ceiling effect within the cohort. Flexion synergy-related biceps EMG was not significantly correlated with Fugl-Meyer, but the measure showed variability across participants and may have been under-powered by the sample size. The results provide evidence in support of the criterion validity of reaching distance at a standardized load and emergence threshold as quantitative measures of flexion synergy. Kinematic and kinetic protocols allow for more direct high-resolution measurement of flexion synergy impairment than observational assessments. The precision of reaching distance at a standardized load and emergence threshold is critical for longitudinal clinical trials and clinical practice seeking to measure changes in flexion synergy expression over time.

Effects of Dry Needling on Spinal Reflexes

Gretchen Seif, Alan Phipps, Blair Dellenbach, Aiko Thompson

Medical University of South Carolina, Charleston, USA

Background: In persons after stroke and other CNS disorders, spasticity can lead to secondary changes in spinal and supraspinal pathways, and together with weakened voluntary control and activation of skeletal muscles, can limit joint motion, impair motor control, and negatively impact function and quality of life. An increasing number of physical therapists worldwide have started administering trigger point deep dry needling (DDN) to treat spasticity and associated pain in people with orthopedic and CNS injuries. Locally, DDN at myofascial trigger points (MTrP) disrupts dysfunctional motor end plates and increases the local blood flow and oxygenation while stimulating nociceptors (sensory afferents). Conceivably, excitation of nociceptors would influence the excitability of multiple spinal pathways. Yet, currently, effects of DDN on spinal somatosensory processing are not well understood. Thus, the first step towards understanding the neurophysiological mechanisms of DDN is investigating the effects of DDN on spinal somatosensory pathways in persons without known neurologic diagnoses and persons with stroke.

Methods: Before and 0, 90 minutes, and 72 hours after DDN of the medial gastrocnemius (MG) MTrP, the H-reflex and M-wave recruitment curves are measured in the soleus, MG and lateral gastrocnemius (LG). Passive ankle range of motion (ROM) is also measured before and after DDN.

To date, 10 adults of 22-57 years old (median 37.5) without known neurological conditions have been enrolled into the study from the university community.

Results: In individuals without known neurological conditions, the MG H-reflex latency was prolonged by 1-3 ms at 0 and 90 minutes post DDN (in 8/10 individuals) and then returned to the pre DDN latency at 72 hours post. Similar short-term prolongation of H-reflex latency was also observed in some of them: 4/10 in the soleus and 2/10 in the LG. In the MG, H-reflex amplitude was not changed at 0 and 90 minutes post but significantly decreased at 72 hours post DDN (p=0.01 by paired t-test) by about 10%. No systematic changes were observed in the amplitudes of non-treated muscles (i.e., LG and soleus, p=0.08-0.47). Ankle dorsiflexion ROM was increased by »4 deg at 0 minutes and 72 hours post DDN (but not at 90 minutes post).

Conclusion: To the author’s knowledge, this is the first study attempting to quantify neurophysiological effects of DDN at the spinal level. The results to date indicate that DDN produces acute short-term effects (as seen in the H-reflex latency shift at 0 and 90 minutes post) and small yet significant long-term effects (as seen in the reduced H-reflex amplitude at 72 hours post) in spinal pathways of the treated MG. In order to examine whether or not these measures of spinal cord plasticity are present in the stroke-impaired CNS, studies of individuals after stroke are currently underway.

A technology-enhanced, comprehensive continuum of stroke recovery care: an exemplar case

Harrison Segall ¹ , Haytey Haaf ² , Amelia Tenberg ¹ , Ken Johnson ³ , Jyo Supnekar ¹ , Stephanie Orient ¹ , Carey Shallal ¹ , Victor Urrutia ¹ , Preeti Raghavan ¹ , Steven Zeiler ¹ , Mona Bahouth ¹

¹Johns Hopkins Hospital, Baltimore, USA. ²Johns Hopkins Hospita, Baltimore, USA. ³Johns Hopkins, Baltimore, USA

Objective: Ideal recovery post-stroke requires efficient care transitions coupled with the right treatments, at the right dose, at the right time. We have developed a technology-enhanced, post-stroke recovery paradigm to achieve this objective. Central to the program are two interactive rehabilitation exercise systems, the hospital-based MindPod technology deployed in an Immersive Treatment Room (MPITR) and the MindMotion^TM GO (MMGO) portable technology for use in the hospital room and at home. Here we describe our pragmatic, interprofessional paradigm of care contextualized by a single case.

Methods: Post-stroke, patients follow a prescribed treatment plan that can be tailored to their deficits. Beginning 24 hours post-stroke, patients are exposed to the MPITR along with in-person therapy. At discharge, patients are enrolled in the Joint Stroke Telemedicine Program (JSTTEP), designed to bridge care from hospital to rehabilitation. At one week post-discharge, patients have a joint telemedicine visit with stroke neurology and physical therapy. At two weeks post-discharge, patients have a joint telemedicine visit with physiatry and occupational therapy. Referrals to the MMGO program are made if patients have substantial deficits that are amenable to synchronous and asynchronous telerehabilitation using the technology.

Results: Ms. S was a 28-year-old with a large MCA stroke treated with intravenous thrombolysis and mechanical thrombectomy. She had global aphasia and right hemiparesis with impaired balance mobility, and ADL’s. In addition to standard therapy, she received three hours of treatment on the MPITR during her hospitalization. After discharge, she received PT, OT, and speech therapy through telemedicine, including 5 synchronous PT and 32 asynchronous MMGO sessions. At 12 weeks post-stroke, her NIHSS score = 3 (original = 13), and AMPAC Mobility T-Score = 74.72(original = 45.55). This represents recovery from a significant neurological deficit and limited mobility, to almost complete neurological recovery and functional independence.

Conclusions: This case demonstrates the benefits of providing high dose early and continuous technology-enhanced rehabilitation for optimal post-stroke recovery. The use of a telemedicine-guided, technology-enriched program has the potential to improve access to care and amplify the benefits of post-stroke treatments.

A study of single task vs dual task training on balance and gait performance in stroke patients

Dr. Hiral gandhi ¹ , Dr. Neha Verma ²

¹INSTITUTE OF NEUROSCIENCES, SURAT, India. ²SPB PHYSIOTHERAPY COLLEGE, SURAT, India

Background: Difficulty with balance and postural control in stroke patients leads to defective movement ability and disables walking and activities of daily living. Therefore, balance is the most basic requirement to prevent falling, undertake the activities of daily living, and maintain an independent life in stroke patients. The ability to walk independently is a prerequisite for many daily activities. Independent walking is a primary goal and a reasonable expectation for most patients. Most often the balance is trained under single-task conditions. Single-task training involves practicing functional tasks requiring balance (eg, standing, walking, and transfer) in isolation, whereas dual-task training allows the coordination of various tasks via the simultaneous performance of 2 or more tasks. Dual task training can be given under different sets of instructions- fixed priority (FP) and variable priority (VP) instructional set. Subsequent work showed poor evidence on effective approaches to training balance and gait under dual-task conditions with different instructional sets in the stroke population. We aimed to address this gap in the literature by using an established dual-task training protocol in stroke patients.

Methodology: 39 stroke patients were allocated in one of the three groups using quasi randomization: single-task training, dual-task training with fixed-priority instructional sets, and dual-task training with variable priority instructional sets. Participants underwent 4 weeks training sessions and were analyzed for balance and gait using Berg Balance Scale and Dynamic Gait Index at baseline and after 4 weeks.

Results: Paired t-test was used to analyze the differences in BBS score and DGI score pre and post training within each group. Results showed that single task training and dual task training both were effective to improve balance and gait performance with p value< 0.0001, highly significant at 95% confidence interval. ANOVA and post hoc analysis was carried out for between group comparison and results shows that dual task training with variable priority instructional sets proved to be most significant one.

Conclusion: Study concluded that single task training and dual task training are effective to improve balance and gait in stroke patients with the dual task training with variable priority instructional sets being the most significant one.

Feasibility and utility of remotely supervised game-based movement priming in chronic stroke: a case series

Hyosok Lim, Sonia Pradhan, Nicholas Marjanovic, Cristian Luciano, Sangeetha Madhavan

University of Illinois at Chicago, Chicago, USA

Background & purpose: Movement-based priming is the use of specific movement paradigms that prepares neural processes for subsequent learning to augment the effects of motor therapy. Priming has been increasingly utilized in laboratory and clinical settings, however the feasibility of this approach at home has not been studied. To administer movement priming at home, we developed a cost-effective game-based technology (DIG-I-PRIME^TM system) that engages the user in repeated skilled movements using the ankle. This study aimed to determine the feasibility and preliminary efficacy of an 8-week telerehabilitation based priming and functional training protocol on walking outcomes and neural mechanisms in chronic stroke survivors.

Methods: Three individuals with chronic stroke completed 24 sessions of a remotely supervised program over 8 weeks monitored via video-conference. Telerehabilitation consisted of 20 minutes of movement priming using the DIG-I-PRIME^TM followed by 40 minutes of functional exercise training focusing on lower limb strength and balance. We evaluated feasibility using frequency of reported adverse events, participant adherence, and participant perception using open feedback questionnaires. Efficacy was assessed by evaluating walking outcomes using the 10-m walk test and Functional Gait Assessment scale. Neurophysiological measures were measured using transcranial magnetic stimulation to assess descending corticomotor excitability of the paretic tibialis anterior muscle. Outcomes were collected at baseline, at 4-week, and after training.

Results: All participants completed the priming and functional exercises at home without any reported adverse events and reported the telerehabilitation program to be ‘enjoyable’, ‘fun to play’, and ‘good challenge’. Post-intervention changes in the 10-m walk test and Functional Gait Assessment exceeded their minimal clinically important difference (range: 0.10-0.31m/s and 4-7 points respectively). Changes in corticomotor excitability of the paretic tibialis anterior muscle demonstrated medium to large effect size for all three participants.

Conclusion: Our results indicate that a remotely supervised game-based movement priming and exercise training program is safe and feasible for stroke survivors to perform at home. In addition, improved walking and neural drive provides preliminary evidence of this intervention to be effective as a telerehabilitation priming strategy for stroke motor recovery. By engaging the participant in their own home using telerehabilitation, we aim to deliver an effective dosage of therapy in addition to broadening rehabilitation access.

Paretic arm motor behavior is related to abnormal state-dependent modulation of interhemispheric inhibition in stroke

Jasmine Mirdamadi, Karla Arevalo-Alas, Liana Kam, Michael Borich

Emory University School of Medicine, Atlanta, USA

Cortical-cortical interactions in motor cortex (M1) (e.g., interhemispheric inhibition (IHI)) are essential for normal motor control. However, the role of IHI in paretic arm motor recovery post-stroke remains controversial. The classical IHI imbalance model posits that reduced inhibition from the ipsilesional (ipsi-M1) to the contralesional hemisphere (contra-M1), hence hypoexcitability in ipsi-M1 and hyperexcitability in contra-M1, impairs motor recovery. However, recent evidence challenges this view, reflecting the heterogeneity of stroke neurophysiology and its relation to behavior. Our understanding of IHI is incomplete due, in part, to limited consideration of state-dependent effects on interhemispheric circuits. Specifically, IHI is typically measured with the paretic arm at rest, which may not represent the role of IHI during movement. Here, we investigated state-dependent modulation of IHI in chronic stroke survivors (N=12) and neurotypical older adults (NOA) (N=10), and characterized relationships to motor behavior. We used dual-coil transcranial magnetic stimulation to measure IHI targeting the first dorsal interosseous (FDI) M1 representation in two states: 1) at rest and 2) during contralateral FDI isometric contraction. IHI was measured by delivering a suprathreshold conditioning stimulus 8-msec (short interval) or 50-msec (long interval) prior to a suprathreshold test stimulus over contralateral M1. IHI was quantified as the ratio of the conditioned motor evoked potential (MEP) amplitude compared to the unconditioned MEP. We assessed paretic arm impairment with the upper extremity Fugl-Meyer Assessment (UEFMA) and strength with the Shoulder Abduction/Finger Extension (SAFE) score. In both groups, we assessed manual dexterity with the Nine-Hole Peg Test. Stroke survivors demonstrated less IHI compared to NOA, with similar IHI magnitudes between ipsi- and contra-M1. Stroke survivors showed less IHI modulation between rest and active states compared to NOA. Individual differences in IHI modulation (active – rest) were related to paretic motor behavior. Greater short interval IHI disinhibition in ipsi-M1 was related to lower UEFMA scores (i.e., greater impairment) (R= -0.728, p= 0.032). Similar associations were observed for long interval IHI modulation in ipsi-M1, though did not reach statistical significance. Greater long interval IHI disinhibition in contra-M1 was related to lower UEFMA and SAFE scores (R= -0.681, p= 0.030; R= -0.656, p= 0.028). In contrast, there were no relationships between resting IHI and paretic motor behavior. Finally, in NOA, there were no associations between IHI and NHPT. Similar levels of ipsi- and contra-M1 IHI counter the classical IHI imbalance model. In contrast to previous findings, more typical state-dependent IHI modulation (i.e., disinhibition) was associated with reduced motor behavior. Therefore, state-dependent compensatory cortical reorganization may support paretic arm motor behavior. Since these relationships were absent at rest, state-dependent interhemispheric interactions could offer biomarkers of functional cortical reorganization that may predict stroke recovery and inform future models of stroke recovery.

Feasibility & Utility of Corticomuscular Coherence Measurement in Early Stroke Motor Recovery: A Preliminary Analysis

Jasper Mark, Rachana Gangwani, Eric Zheng, Rachel Vaughn, Jessica Cassidy

University of North Carolina at Chapel Hill, Chapel Hill, USA

Introduction: The heterogeneity of stroke has propelled the development of stroke recovery biomarkers to enhance clinical outcome and treatment response prediction. Corticomuscular coherence (CMC) is a measure of synchrony between brain and muscle signals as captured by electroencephalography (EEG) and electromyography (EMG). As a measure of communication between central and peripheral nervous systems, CMC may provide valuable insight to post-stroke motor recovery that resting-state neuroimaging biomarkers cannot convey. This ongoing study sought to [1] confirm the feasibility of CMC acquisition in patients with stroke in an inpatient rehabilitation facility (IRF) and [2] examine the utility of CMC measures in early stroke motor recovery.

Methods: Participants completed two research visits consisting of behavioral (Upper Extremity Fugl-Meyer, UEFM; Action Research Arm Test, ARAT) and CMC testing occurring near the time of IRF admission and discharge. CMC testing encompassed simultaneous collection of functional brain activity using a 256-lead EEG system and muscle activity with EMG leads on bilateral biceps brachii, flexor and extensor digitorum (FD, ED), and first dorsal interossei (FDI) muscles. During CMC testing for each extremity, participants performed an isometric grip task at 20% of their maximum for 2 blocks of 40 trials with visual feedback of force output provided. Compensatory movements were monitored throughout testing with EMG and visual observation. CMC measurements between leads overlying ipsilesional primary motor cortex (iM1) and the above muscles were computed across alpha-mu (11-14 Hz), low beta (13-19Hz), and high beta (20-30Hz) frequency bands. CMC change during hospitalization and associations between CMC and behavioral change were examined using non-parametric statistics.

Results: Of the 6 individuals enrolled, 4 (2 female, aged 63±6.2 years) with ischemic and hemorrhagic stroke depicting moderate-severe motor impairment (baseline UEFM=23.3±24.0) and function (baseline ARAT=18.5±23.2) completed study procedures during their hospitalization (average stay=17.2±8 days). Individuals demonstrated motor improvement during hospitalization (UEFM change=9.5±3.9, ARAT change=7.5±15.7). Two participants were not able to physically complete CMC testing on their affected side at the initial visit but did so at discharge. An average of 26 (visit 1) and 25 (visit 2) trials were retained for analyses. Participants typically demonstrated trends of decreased CMC between iM1 and FD and ED muscles across all frequency bands and increased CMC between iM1 and FDI in the alpha-mu frequency band. The latter showed a positive correlation trend with ARAT change (ρ=0.9487, p=0.05).

Conclusion: Preliminary findings of this ongoing study support the feasibility of CMC measurement in the IRF. Despite our small sample, trends of frequency and muscle specific CMC modulation during hospitalization may indicate improved nervous system communication that may relate to early motor recovery. We plan to confirm these initial findings with additional participants and a 90-day post-stroke follow-up visit.

Predicting clinically significant improvement after robot-assisted upper limb rehabilitation in subacute and chronic stroke

Jae Joon Lee ¹ , Joon-Ho Shin ^1,2

¹Department of Rehabilitation Medicine, National Rehabilitation Center, Seoul, Korea, Republic of Korea. ²Translational Research Center for Rehabilitation Robots, National Rehabilitation Center, Seoul, Korea, Republic of Korea

Background: Prior studies examining predictors for favorable clinical outcomes after upper limb robot-assisted therapy (RT) have shortcomings.

Objective: To identify meaningful predictors and a prediction model for clinically significant motor improvement in upper limb impairment after RT for each time phase.

Methods: This retrospective, single-center study enrolled patients with stroke who received RT using InMotion2 along with conventional therapy (CT) from January 2015 to September 2019. Demographic characteristics, clinical measures, and robotic kinematic measures were evaluated. The primary outcome measure was the Fugl-Meyer Assessment Upper Extremity (FMA-UE) and we classified patients who obtained improvement more than the minimal clinically important difference as responders for each time phase. Univariable and multivariable logistic regression analyses were performed to assess the relationship between potential predictors and RT responders and determine meaningful predictors. Subsequently, meaningful predictors were included in the final prediction model.

Results: One hundred forty-four patients were enrolled. The Hand Movement Scale and time since onset were significant predictors of clinically significant improvement in upper limb impairment (P=0.045 and 0.043, respectively), as represented by the FMA-UE score after RT along with CT in patients with subacute stroke. These variables were also meaningful predictors with borderline statistical significance in patients with chronic stroke (P=0.076 and 0.066, respectively).

Conclusions: Better hand movement and a shorter time since onset can be used as realistic predictors of clinically significant motor improvement in upper limb impairment after RT with InMotion2 alongside CT in patients with subacute and chronic stroke. This information may help healthcare professionals discern optimal patients for RT and accurately inform patients or caregivers about outcomes after RT.

Quantifying the Impact of Hemiparetic Stroke on Trunk Motor Control During Reaching

Kathleen Suvada, Jasjit Deol, Julius Dewald, Ana Maria Acosta

Northwestern University, Chicago, USA

The trunk provides a stable base of support and facilitates interaction of the limbs with the environment. Post hemiparetic stroke, damage to descending corticospinal pathways impairs typical trunk and arm motor control, therefore impacting activities of daily life. In the arm, impairments include weakness, hyperactive stretch reflexes and involuntary coupling of shoulder abduction with elbow, wrist, and finger flexion, or flexion synergy. Based on previous findings, we believe the flexion synergy is a result of increased reliance on reticulospinal pathways, also responsible for maintenance of posture. In the trunk, studies found excessive trunk movement while reaching with the paretic limb, larger sway area during quiet sitting, weakness, and altered coordination compared to controls. However, little is known about the impact on reaching and if trunk function is impacted by the flexion synergy. The goal of this study is to measure the effect of trunk impairments arising from a stroke on reaching ability and to understand the consequences of increased use of the reticulospinal system wherein the arm and trunk are utilizing the same system. We designed a novel experimental paradigm that uses a haptic robotic device to create virtual reaching environments while measuring trunk and arm kinematics with a motion capture system, primary trunk and arm musculature activity via surface EMG, trunk center of pressure with flexible pressure mats placed on the seat and back.

Two controls and three stroke participants provided informed consent prior to enrolling in the study. Participants were seated with their pelvis restrained and the back of the chair in place or reclined. The paretic arm was coupled to a haptic device that allows motion on a horizontal plane and can apply vertical loads. Participants were instructed to reach as far as possible on either a virtual table or against 50% of their maximum shoulder abduction torque (SABT) while also maintaining trunk posture during unrestrained trials. Preliminary results show that in stroke participants, reaching distance was greatest when the trunk was restrained while reaching on the table, whereas reaching distance was the shortest when the trunk was unrestrained and the arm was generating 50% maximum SABT. Additionally, we observed greater activation of trunk flexors when the trunk was restrained. These preliminary results exemplify the importance trunk function in the understanding of reaching deficits post stroke as the arm is not functioning in isolation. Further testing will allow us to determine whether the flexion synergy also impacts the trunk, and its effect on reaching function.

Modified Rankin Scale Does Not Capture Precise Recovery Phenotypes after Acute Stroke

Kimberly Erler ^1,2 , Rui Wu ² , Julie DiCarlo ² , Leigh Hochberg ^2,3,4 , Steven Kautz ^5,6 , Lee Schwamm ² , Steven Cramer ^7,8 , Seth Finklestein ² , David Lin ^2,3

¹MGH Institute of Health Professions, Boston, MA, USA. ²Massachusetts General Hospital, Boston, MA, USA. ³Department of VA Medical Center, Providence, RI, USA. ⁴Brown University, Providence, RI, USA. ⁵Medical University of South Carolina, Charleston, SC, USA. ⁶Ralph H Johnson VA Medical Center, Charleston, SC, USA. ⁷University of California, Los Angeles, Los Angeles, CA, USA. ⁸California Rehabilitation Institute, Los Angeles, CA, USA

Background: The Modified Rankin Scale (mRS) is a global disability measure that is widely accepted as the gold standard for capturing outcomes in acute stroke clinical trials, but it lacks granularity. Impairment- and function-based outcomes have the potential to capture phenotypes more precisely and to measure change from post-stroke baseline. This study aimed to (1) identify relationships that the mRS has with commonly used impairment and function outcomes, (2) determine whether mRS levels capture distinct phenotypes of impairment and function, and (3) compare mRS outcomes to clinically meaningful changes in impairment and function.

Methods: Patients with upper extremity weakness after ischemic stroke were serially assessed (during acute stroke hospitalization and 90-days post-stroke) with a battery of impairment- and function-based outcome measures including Fugl-Meyer Upper Extremity Motor (FMA-UE), Grip Strength (Grip), Box and Blocks (BBT), Nine-Hole Peg (9-HP), Gait Velocity (GV), Timed Up and Go (TUG), National Institutes of Health Stroke Impact Scale (NIHSS), and Barthel Index (BI). The Stroke Impact Scale -16 (SIS-16) and mRS were assessed at day 90. Spearman Rho correlations were performed to assess the relationship between the 90-day mRS and the function and impairment outcomes. Separate Kruskal-Wallis H tests with post hoc comparisons were performed with mRS as the independent variable and other outcomes as dependent variables. Clinically important changes in UE impairment, independence with activities of daily living, and mobility from post-stroke baseline to 90-days were examined in relation to 90-day mRS.

Results: N = 73 patients completed testing. While the 90-day mRS was highly correlated with all function and impairment outcomes (p<0.001), within each mRS level, median and interquartile ranges of individual function and impairment outcomes revealed substantial variation. Post-hoc tests on adjacent mRS levels revealed that mRS levels 1 and 2 did not differ in relation to any of the outcome measures; mRS levels 2 and 3 only differed on SIS, NHP, and BBT; and mRS levels 3 and 4 differed on NIHSS, TUG, BI, GV, and NHP. Notably, there were no differences on UE impairment focused measures (i.e. FMA-UE and GS) between any adjacent levels of the mRS. There were substantial numbers of patients who experienced clinically meaningful changes in impairment and function in the first 90 days post-stroke who did not achieve good mRS outcome.

Conclusion: Our results provide insights into what the mRS captures well and what it does not. While the mRS is broadly related to domain-specific impairment and global functional outcomes, mRS levels do not capture distinct impairment and function phenotypes or clinically meaningful changes from post-stroke baseline. Progress in the field of stroke recovery requires the optimal selection and deployment of outcome measures that reflect neurological and functional status and are meaningful to patients and caregivers.

Pathological inhibition limits motor unit rate modulation during voluntary contractions in a muscle-dependent manner post-stroke

Laura McPherson ¹ , Francesco Negro ² , Christopher Thompson ³ , Keith Lohse ⁴ , Randall Powers ⁵ , Dario Farina ⁶ , CJ Heckman ⁷ , Jules Dewald ⁷

¹Washington University, St. Louis, USA. ²Universita degli Studi di Brescia, Brescia, Italy. ³Temple University, Philadelphia, USA. ⁴University of Utah, Salt Lake City, USA. ⁵University of Washington, Seattle, USA. ⁶Imperial College London, London, United Kingdom. ⁷Northwestern University, Chicago, USA

Muscle weakness in the arm of individuals with chronic hemiparetic stroke is hallmarked by a decreasing proximal to distal gradient of severity whereby the hand is the most impaired. A few studies have examined motor unit (MU) control of force post-stroke in isolated muscles, demonstrating that firing rates are low and rate modulation is reduced. However, available studies do little help understand post-stroke weakness in proximal vs. distal muscles because none have recorded MU discharge from different muscles within the same cohort of participants. Thus, whether deficits in MU rate modulation are commensurate with the gradient of post-stroke weakness in the arm is an open question.

We used high-density surface EMG decomposition to quantify the mean discharge rate vs. joint torque relationship in MU of the deltoid, biceps, and finger flexors during steady contractions at multiple torque levels between 10% and 40% of maximum. We collected data from 15 individuals with chronic moderate-to-severe post-stroke hemiparesis and 10 age-matched neurologically-intact individuals. We used a linear mixed effects regression model (muscle x group x torque + muscle x group x torque²) to test for differences in rate modulation among muscles within and between groups.

Our primary finding is that MU rate modulation in the stroke group was unchanged in the deltoid, substantially impaired in the biceps, and absent in the finger flexors. Rate modulation increased when comparing muscles from proximal to distal in the control group but decreased for the same comparison in the stroke group. For the paretic biceps and finger flexors, the extent of impaired rate modulation was correlated with clinical motor impairment. These findings were not due to strength differences between groups.

Previous simulation work using motoneuron models demonstrated that the pattern of MU rate modulation resulting from increasing activation of the motoneuron is reflective of both intrinsic motoneuron excitability and the temporal pattern of inhibitory inputs to the motoneuron. To help interpret our findings, we used these models to simulate MU firing rates in response to steady contractions of increasing intensity. The modelled firing rate pattern most consistent with our experimental findings from the post-stroke biceps and finger flexors was the one associated with inhibitory inputs that scaled proportionally with excitation of the motoneuron. When these inputs were combined with a high level of intrinsic motoneuron excitability, the modelled MU firing pattern was consistent with the absent rate modulation of the post-stroke finger flexors.

Despite the common assumption that weakness post-stroke is due exclusively to decreased excitation of the motoneuron pool as a result of corticospinal damage, our findings suggest that, rather, pathologic inhibition combined with high intrinsic motoneuron excitability appears to substantially limit motor output, revealing a specific mechanism that can potentially be targeted with multi-modal rehabilitation interventions.

Video Survey of Occupational Therapy Exercises Sessions During Acute Rehabilitation for Stroke

Madeline White ¹ , Noah Balestra ¹ , Mouhamed Diakhate ¹ , Linda Riek ² , Ania Busza ¹

¹University of Rochester, Rochester, USA. ²Nazareth College, Rochester, USA

Objective: To perform a survey of upper extremity (UE) exercises performed by patients with recent stroke in our inpatient rehabilitation gym in order to count number of exercise repetitions, identify active and inactive time, and assess the relative use of the affected and unaffected limbs during therapy sessions.

Background: Both animal and human studies suggest that repetitive exercises improve motor outcomes after stroke, however which features of an exercise regimen are critical for optimizing recovery remain to be determined. Our lab is using wearable sensors and machine learning to develop a system to classify and quantify therapy exercise repetitions performed during the inpatient rehabilitation period. As part of this study, we are performing a video survey of occupational therapy sessions occurring in our inpatient rehabilitation gym.

Design: Individuals with UE weakness due to recent (<4 weeks) unilateral stroke admitted to our inpatient rehabilitation unit gave permission to have their routine clinical occupational therapy sessions video recorded. The video data set was then reviewed by 2 independent study investigators and the activity for each frame was labeled to identify the exercise/activity being performed, active and passive arms, level of difficulty for the patient, type of exercise (active, passive, functional-realistic, and functional-unrealistic), and number of repetitions. The total duration of each exercise was calculated as well as the amount of inactive time spent not performing any exercise.

Preliminary Results: In this ongoing study we have thus far reviewed 16 occupational therapy sessions (total of 7.7 hours) looking at therapy sessions from 14 subjects. The mean number of repetitions per session was 101.5 with a standard deviation of 117.4, and the range was 4-500. The mean number of repetitions per minute was 4 with a standard deviation of 5.1, and a range of .6-22 repetitions per minute. The mean percent of the session active was 54.6% with a standard deviation of 18.2%, and the mean time spent resting was 45.4%. Of note, the mean percent of the session performing a bilateral exercise was 47.5% with a standard deviation of 36.9%, while 52.5% of the session subjects were engaged in unilateral tasks using their affected arm.

Conclusions/Future Direction: Similar to prior studies, we see a wide range of number of repetitions performed during a standard rehabilitation session, indicating that time spent in therapy is a poor indicator of rehabilitation dose. We also find roughly half of the exercises were unilateral tasks using the affected arm. Future work will investigate the relative roles the affected and unaffected arms during bilateral exercise activities.

Validity and usability of a wearable, multi-sensor system for monitoring upper and lower limb activity in chronic stroke survivors in a community setting

Justin Rowe ¹ , Marika Demers ² , Lauri Bishop ² , Daniel Zondervan ¹ , Carolee Winstein ²

¹Flint Rehabilitation Devices, Irvine, USA. ²University of Southern California, Los Angeles, USA

Introduction: One of the most impenetrable problems after stroke is the gap between the motor capacity a stroke survivor regains and the amount of motor activity performed in the community.^1,2 While there are useful tools for measuring upper (UL) and lower limb (LL) motor capacity in a clinical setting, few systems exist for accurately measuring UL and LL activity in an unsupervised community setting, especially for individuals with motor impairments.³ The usability of such a system must also be optimized to ensure it is acceptable for extended community use. The goal of this research is to develop a wearable, multi-sensor system that accurately measures natural UL and LL activity, while leveraging end-user feedback to optimize usability.

Methods: During a single visit, 20 chronic stroke survivors expressing mild-to-severe motor impairments donned the MiGo system (Flint Rehab). MiGo consists of five activity monitors worn bilaterally at the wrists and ankles and at the hip on the more affected size. Participants performed the Chedoke Arm and Hand Activity Inventory-7 (CAHAI) and a 2-minute walk test (2MWT). For the CAHAI, MiGo measured the ratio of active movement time to total duration for each arm. For the 2MWT, MiGo measured step counts and stance time symmetry using data from just the hip sensor, one ankle sensor, and both ankle sensors. Agreement between the system and gold-standards (annotated video recordings for CAHAI and validated APDM wearable sensors for 2MWT) was examined. Social acceptability, ease of use, and time to don/doff each sensor were also assessed via end-user feedback.

Results: For the UL, the average active time ratio agreement was 99% for both arms. For the LL, the average step count agreements were 98.95% between the MiGo-hip and the non-paretic MiGo-ankle, 98.4% between the two MiGo-ankles, 75.2% between the APDM and the non-paretic MiGo-ankle, and 68.49% between the APDM and the MiGo-hip. The APDM system counted fewer steps during turns and at slow speeds. The average stance time symmetry agreement was 99% when using either one or both ankle sensors. Social acceptability and perceived usability were high, with 95% of participants indicating a willingness to use all 5 sensors daily. However, 10% of participants noted the sensors were difficult to don and doff. The hip was the most preferred sensor placement whereas the ankles were the least preferred.

Conclusion: The wearable, multi-sensor system accurately measured active UL movement time and stance time symmetry with high perceived usability. For step counts, an alternative gold standard such as annotated video recordings should be used in future analysis, as the APDM sensors used here were not optimized to count steps during turns or for individuals with slower gait, both of which are natural LL activities performed in community settings.

Post-stroke vascular repair and remodeling are facilitated by reactive astrocytes

Michael Williamson, Cathleen Joy Fuertes, Andrew Dunn, Michael Drew, Theresa Jones

University of Texas at Austin, Austin, USA

Reactive astrocytes are a hallmark of central nervous system injury and disease. While reactive astrocytes have beneficial protective functions, they have been viewed as a barrier to neural repair. Through analyzing a dataset on gene expression changes in astrocytes in a mouse stroke model, we identified that reactive astrocytes substantially upregulate an array of genes relevant for neurovascular unit repair and remodeling during the first week post-stroke. Subsequently, we examined whether reactive astrocytes interact with newly formed blood vessels in peri-infarct cortex following photothrombotic infarcts in mice. Using both longitudinal in vivo imaging and post-mortem immunohistochemical approaches, we found that astrocytic processes surround new vessels. These findings led us to hypothesize that reactive astrocytes are involved in regulating post-stroke vascular reparative responses. We next evaluated the consequences of ablating a subset of reactive astrocytes surrounding photothrombotic strokes using glial fibrillary acidic protein-thymidine kinase (GFAP-TK) mice. GFAP-TK mice permit conditional ablation of proliferating astrocytes by phosphorylating exogenous ganciclovir, administered during the first week after stroke, to form a cytotoxic nucleotide analogue that arrests DNA replication and induces apoptosis. Mice with ablated reactive astrocytes showed marked dysfunction in vascular repair and remodeling relative to controls, characterized by: a sparse peri-infarct vascular network, diminished neovascularization, reduced blood flow, increased vascular leakage, and sparse coverage of vessels by basement membrane, pericytes, and glycocalyx. Moreover, mice with ablated astrocytes showed impaired motor recovery measured with cylinder and grid walking tasks. These findings position reactive astrocytes as critical mediators of vascular repair and remodeling after stroke. More generally, our findings show that reactive astrocytes can have pro-reparative functions following nervous system damage.

Examining the relationship between motor control and abnormal synergies during arm and index finger movement in chronic stroke patients

Myriam Taga ¹ , Yoon N. G. Hong ² , Charalambos C. Charalambous ^3,3 , Sharmila Raju ¹ , Jing Lin ¹ , Elisa Stern ¹ , Pietro Mazzoni ⁴ , Jinsook Roh ² , Heidi M. Schambra ¹

¹Department of Neurology, NYU Langone, School of Medicine, New York, USA. ²Department of Biomedical Engineering, University of Houston, Houston, USA. ³Department of Basic and Clinical Sciences, Medical School, University of Nicosia, Nicosia, Cyprus, Greece. ⁴Department of Neurology, Washington University, School of Medicine in St. Louis, St. Louis, USA

Introduction: With the corticospinal tract (CST), the corticoreticulospinal tract (CReST) is a major descending motor pathway with widespread bilateral innervation. In animals, CST damage causes a loss of motor control and prompts reorganization in the CReST, possibly with stronger connectivity to arm flexors (e.g. biceps (BIC)) than finger abductors (e.g. first dorsal interosseous (FDI)). CReST reorganization may also contribute to widespread muscle co-activations (i.e. abnormal synergy expression) in the paretic upper extremity (UE). Here, we posited that CReST reorganization after stroke targets the BIC more than the FDI in humans. We predicted that CReST activity, manifesting as abnormal synergy expression, would be more strongly evoked by skilled arm flexion than finger abduction in stroke patients.

Methods: We studied the paretic UE of 14 chronic stroke patients (F: 8; mean age: 64 (44-85) years; mean post-stroke time: 5 (0.5-14.4) years) and the matched UE of 14 healthy controls (F: 6; mean age: 55 (36-81) years). Subjects used their arm or index finger to move an onscreen cursor through an arc-shaped channel while the remainder of the UE was restrained. We recorded effector kinematics with an infrared camera and electromyographic (EMG) signals from triceps (TRI), deltoid (DLT), BIC, extensor digitorum, flexor carpi radialis (FCR), flexor digitorum superficialis (FDS), and FDI. To quantify movement error, we calculated the average radial distance between the cursor path and the outer channel edge. To quantify abnormal muscle synergies, we applied a non-negative matrix factorization algorithm to the EMG data to identify muscle synergies and calculated the similarity of the synergy vectors between patients and controls; higher similarity scores indicate more normal synergy patterns. We calculated muscle co-activations using correlations between EMG signals of each muscle-pair. We examined group differences with independent t-tests and control-synergy relationships with correlations.

Results: Movement errors were higher in patients than controls for the arm (p<0.01) and trended higher for the finger (p=0.074). In the arm, movement errors were inversely related to synergy similarity scores (p<0.01). Higher errors also related to greater FDI-FCR, BIC-TRI, BIC-DLT, and TRI-DLT co-activation (all p<0.05). In the finger, movement errors were unrelated to synergy similarity scores. Lower movement errors related to greater FDS-TRI co-activation (p<0.05).

Discussion: In the arm, we found that as motor control worsened, the expression of abnormal synergies increased, indicating that CReST activation may increase with loss of CST function. Muscle co-activation was widespread in the UE, in keeping with CReST’s multilevel spinal branching. We did not find a relationship between motor control and synergy expression with finger movement, although the long-range co-contraction between the FDS and TRI may speak to a CST-driven stabilizing strategy. Our findings strengthen the notion that CReST reorganization after stroke may preferentially target the arm flexor and its synergies.

Corticoreticulospinal tract neurophysiology in healthy and chronic stroke subjects

Myriam Taga ¹ , Charalambos C. Charalambous ^2,3 , Sharmila Raju ¹ , Jing Lin ¹ , Elisa Stern ¹ , Heidi M. Schambra ¹

¹Department of Neurology, NYU Langone, School of Medicine, New York, USA. ²Department of Basic and Clinical Sciences, medical School, University of Nicosia, Nicosia, Cyprus, Greece. ³Center for Neuroscience and Integrative Brain Research (CENIBRE), Medical School, University of Nicosia, Nicosia, Cyprus, Greece

Background: The corticoreticulospinal tract (CReST) is a major descending motor pathway in humans, but little is known about its relative innervation of proximal versus distal upper extremity (UE) muscles. In addition, CReST is believed to reorganize after corticospinal injury, but changes in its projections to different paretic muscles remain unknown. Here, we used transcranial magnetic stimulation (TMS) to probe the functional connectivity of the contralesional CReST to an arm muscle (biceps (BIC)) and an intrinsic hand muscle (first dorsal interosseous (FDI)) in healthy and stroke subjects.

Methods: In this cross-sectional observational study, we examined 15 healthy (F: 7; mean age: 54 (44-81) years; mean UE Fugl-Meyer Assessment (FMA) score: 65 (63-66)) and 16 chronic stroke subjects (F: 10; mean age 62 (44-85) years; mean UE FMA score: 49 (23-64); mean time since stroke: 5 (0.5-14.4) years). We applied TMS to the contralesional hemisphere (assigned in healthy subjects) to elicit ipsilateral motor evoked potentials (iMEPs). We measured contralesional CReST functional connectivity (iMEP presence/absence) and projection strength (iMEP size; mV*ms) to the paretic BIC and FDI. We also measured paretic muscle maximum voluntary contraction and segmental FMA subscores. We examined differences in CReST projections between muscles and subject groups using Fisher’s exact tests and general linear mixed models, and examined neurophysiological-behavioral relationships with Pearson’s and Spearman’s correlations.

Results: The contralesional CReST made functional connections to both muscles of most subjects (iMEP presence/absence: healthy BIC 14/1, healthy FDI 15/0; stroke BIC 11/5, stroke FDI 15/1). CReST functional connectivity did not differ between muscles in either healthy or stroke subjects (all p>0.172), and did not differ between subject groups for either muscle (all p=1.0). However, CReST projection strength for the muscles diverged between subject groups, manifesting as larger iMEPs in FDIs than BICs in healthy subjects (1.9 mV*ms, p=0.042) and larger iMEPs in BICs than FDIs in stroke subjects (1.0 mV*ms, p=0.042). Muscle iMEP sizes did not significantly differ between healthy and stroke subjects. Muscle strength related to iMEP size in only the paretic BIC of stroke subjects (r(6)=0.853, p=0.007). There was no relationship between FMA subscores and iMEP size for either muscle in either subject group.

Conclusion: Our findings indicate that the contralesional CReST has readily identifiable connections to the paretic BIC and FDI. In healthy subjects, the identification of a stronger CReST projection strength to the FDI challenges the notion of a proximal innervation bias by the reticulospinal tract. The shift in projection strength to the BIC after stroke reinforces the concept that the CReST reorganizes after CST injury, with circumscribed behavioral relevance. To confirm a recovery role of the CReST, a longitudinal observation of recovering behavior relating to changing CReST neurophysiology is required.

Pre-morbid Ischemic Stroke Functioning Impacts Stroke Recovery, Independent of Stroke Severity

Royce Kwon ¹ , Rebecca F. Gottesman ² , Michelle C. Johansen ²

¹University of Hawai'i at Mānoa, John A. Burns School of Medicine, Honolulu, USA. ²Department of Neurology, The Johns Hopkins University School of Medicine, Baltimore, USA

Background: Although characteristics of an ischemic stroke (IS), such as stroke severity, are known to influence post-stroke recovery, the degree to which pre-morbid functioning accounts for post-IS recovery remains unknown. Similar to how the modified Rankin scale (mRS) is used in defining therapeutic success in IS clinical trials, the impact of pre-morbid functioning on post-IS outcomes could be used as one measures for determining post-IS recovery.

Methods: 167 IS patients (≥18yo) admitted to Johns Hopkins Hospital as part of an ongoing prospective study provided informed consent. Pre-morbid function was defined as activities of daily living (ADL) and instrumental activities of daily living (IADL) prior to hospital admission. ADL and IADL statuses were obtained from physical/occupational/speech-language therapy assessments and social work evaluations. ADL and IADL statuses were assigned a score, 0-6 and 0-7 respectively, with a large score corresponding to a worsening in function. Patient stroke severity and post-IS function were defined as NIHSS and 90-day mRS respectively. Covariates included age (years), black race, sex, body mass index (kg/m²), low density lipoprotein (mg/dL), hemoglobin A1C (%), systolic blood pressure (mmHg), smoking, current drug user, mood disorder, and mild cognitive impairment. Multivariable linear regression models were used to estimate the association between NIHSS (dependent variable) and ADL and IADL, each in separate models, with adjustment for covariates. Multivariable binomial (1) or ordinal (2) logistic regression models were used to determine the association between mRS [(1) 0-2 vs ≥3; (2) 0-6] and ADL and IADL, each in separate models, with adjustments.

Results: 155 patients had NIHSS and 115 had mRS available for analysis. Participants were on average 61yo, black (59%), males (57%). For every 1-point worsening in ADL, there was an increase in NIHSS (β 0.86, 95%CI 0.23-1.48) and a reduced odds of a low (favorable) mRS (OR 0.63, 95%CI 0.43-0.92). With ADL worsening, the odds of a 1 point worse mRS was 1.52 times greater independent of demographics and vascular risk factors (95%CI 1.14-2.02). Worsening IADL was associated with an increase in NIHSS scores (β 0.70, 95%CI 0.16-1.24), but no association was found with mRS. NIHSS was independently associated with a worse mRS (β 0.08, 95%CI 0.01-0.16), but lost significance when covarying for ADL (β 0.07, 95%CI -0.001-0.15), however the effect of ADLs on mRS remained significant, even after adjustment for NIHSS.

Conclusion: Long-term outcomes after IS are dependent on prior level of functioning (ADL), independent of IS severity. The pre-morbid functional status (ADL) should therefore be considered when using measures such as the mRS as indicators for therapeutic success or failure.

Virtual Reality based rehabilitation improves self-perceived quantity and quality of upper limb use after stroke: A systematic review and meta-analysis.

Sandeep Subramanian ¹ , Eder Dominguez ¹ , Danielle Hildom ¹ , Charles Vu ¹ , Melissa Whitmann ¹ , Anjali Sivaramakrishnan ¹ , Joyce Fung ^2,3

¹UT Health San Antonio, San Antonio, USA. ²McGill University, Montreal, Canada. ³Jewish Rehabilitation Hospital Site of Centre for Interdisciplinary Research in Rehabilitation of Greater Montreal, Laval, Canada

Background: Upper limb (UL) hemiparesis is frequently a disabling consequence of stroke. The ability to improve UL functioning is associated with motor relearning. Virtual Reality (VR) technology can be incorporated as a technical platform for application of motor learning principles to deliver personalized interventions. The effects of VR-based rehabilitation on UL motor improvements after stroke have primarily been assessed using motor impairment and activity limitation outcomes. Comparatively less attention has been paid to the effects of the intervention on self-perceived use of the UL in performance of daily life activities. Information on self-perceived use provides insight into real life use of the more-affected UL in daily activities. The study objective was to conduct a systematic review and meta-analyses to examine the influence of VR-based interventions on the self-perceived quantity and quality of UL use after a stroke.

Methods: Using standard methodology, we conducted a literature search of databases including MEDLINE and Google Scholar using keywords related to stroke and VR. We included studies i) involving adult humans with a diagnosis of stroke; ii) using VR-based interventions for UL motor improvement and iii) assessing self-perceived UL use (quantity and quality) with Motor Activity Log (MAL) as the outcome. Studies a) examining the influence of VR-based interventions on the pediatric population and b) single-case studies were excluded. The Downs and Black checklist helped assess quality of the retrieved studies. We compared changes in self-perceived levels of quantity and quality of UL use immediately after the intervention and at retention using standardized mean differences and summary effect sizes. We used the RevMan5 software for conducting the meta-analyses.

Results: We retrieved a total of 15 studies. Quality of the published studies ranked from fair to excellent. Immediately, after completion of the VR intervention, an increase in self-perceived quality of UL use was reported (MAL Quality of movement scores; 15 studies; effect size: 0.36, 95% CI: 0.1 - 0.61, p = 0.006). This positive gain is retained at follow-up from one to six months (11 studies; effect size: 0.73, 95% CI: 0.01 - 1.46, p = 0.049). Similar immediate positive effects were reported for self-perceived quantity (MAL quantity of use scores: 14 studies, effect size: 0.32, 95% CI: 0.04 to 0.59, p = 0.02) as well. The positive gain in quantity of UL use was again retained at follow up from one to six months (11 studies, effect size: 0.57, 95% CI: 0.12 - 1.03, p = 0.01).

Conclusion: VR-based rehabilitation is beneficial in improving self-perceived quality and quantity of post-stroke UL use for performance of daily life activities in the real-world setting.

Can kinematic outcomes help distinguish between different levels of upper limb motor impairment severity?

Sandeep Subramanian ¹ , Lori Hoffman ¹ , Ashleigh Morgan ¹ , Lauren Slagle ¹ , Kelli Westlund ¹ , Melanie Banina ² , Mindy Levin ^3,2

¹UT Health San Antonio, San Antonio, USA. ²Jewish Rehabilitation Hospital site of Centre for Interdisciplinary Research in Rehabilitation of Greater Montreal, Laval, Canada. ³McGill University, Montreal, Canada

Background: Measurement of the quality of upper limb movements provides important complementary information about motor impairment. Recent guidelines have stressed the need to include measures of movement quality as outcomes in individuals with post-stroke hemiparesis. These outcomes can be obtained using kinematic analysis, that provides information on endpoint performance (movement straightness, smoothness, precision, and speed) and movement patterns (joint ranges of motion, trunk displacement). Movement quality outcomes are reliable and valid measures of upper limb motor impairment. However, it is unknown whether these measures, aside from trunk displacement, can help distinguish between levels of motor impairment severity in individuals who have sustained a stroke

Methods: We conducted a retrospective analysis of data from 72 participants collected in four previous studies. All subjects performed unrestrained reaching tasks to a target placed in front of them. Upper limb Fugl-Meyer Assessment (FMA) scores and kinematic data (sagittal trunk displacement, shoulder flexion, shoulder horizontal adduction, elbow extension and movement curvature values expressed as an index of curvature where 1.0 corresponds to a straight line) were considered for analysis. Participants were divided into two groups based on the FMA scores (mild: ≥50/66; moderate-to-severe: ≤49/66). Logistic regression analysis was used to estimate whether the kinematic measures distinguished between different levels of impairment severity. Receiver Operating Characteristic curve analyses estimated sensitivity and specificity. Sensitivity-specificity decision plots provided cutoff values for transition of individuals between groups.

Results: Trunk displacement, shoulder horizontal adduction, elbow extension and movement straightness values discriminated between different levels of upper limb motor impairment severity. A one-unit increase in trunk displacement and movement curvature index values increased the odds of moving from the mild to the moderate-to-severe group by 98% and 2% respectively. A one-unit increase in shoulder horizontal adduction and elbow extension increased the odds of moving from the moderate-to-severe to mild group by 104% and 105% respectively. Cutoff points for transition between groups were as follows: Trunk displacement: 6.5 cm, Shoulder horizontal adduction: 54°, elbow extension: 126° and movement curvature index: 1.2.

Conclusion: Movement quality kinematic variables can help distinguish between different levels of upper limb motor impairment severity. This information can help further justify their use as outcome measures in rehabilitation studies.

Investigating Speed-Dependent Changes in Gait Kinematics: A Comparison between Age-Matched and Post-Stroke Gait

Sarah Kettlety ¹ , Catherine Broderick ¹ , James Finley ¹ , Darcy Reisman ² , Kristan Leech ¹

¹University of Southern California, Los Angeles, USA. ²University of Delaware, Newark, USA

Kinematic gait deviations are common post-stroke and are associated with decreased gait speed,¹ increased metabolic cost,² and increased fall risk.³ Clinical practice guidelines recommend high-intensity gait training at fast speeds for gait rehabilitation for individuals post-stroke to address activity limitations.⁴ Importantly, walking at fast speeds can also improve specific kinematic metrics, without increasing compensatory movements in people post-stroke.⁵ However, it is unclear if these speed-dependent improvements result in gait kinematics that are similar to age-matched controls walking at comparable speeds. The purpose of this study is to determine if walking at faster speeds changes the magnitude of kinematic differences between people post-stroke and age-matched controls.

We hypothesize that this difference will increase as speed increases. We performed a secondary analysis of three datasets from cross-sectional studies of people post-stroke and able-bodied older adults.^5–7 The stroke dataset included lower extremity kinematic data from the paretic side of twenty-eight people > six-months post-stroke (61.79 + 10.52 years, Fugl-Meyer Lower Extremity: 24.04 + 5.82) who walked on a treadmill at four different speeds.^5,6 The control dataset included data from twenty-six age-matched able-bodied adults (62.05 + 8.28 years) who walked on a treadmill at either four⁶ or eight speeds.⁷ We analyzed the right lower extremity kinematic data of each control participant at the four speeds that most closely matched the speeds in the post-stroke dataset. Linear mixed-effects models, with an a priori alpha = 0.05, were used to evaluate the effect of speed and group on circumduction, hip hiking (HH), step length asymmetry (SLA), trailing limb angle (TLA), and peak knee flexion angle (pKF). Random intercepts were included in all models to account for individual differences.

We found that people post-stroke had greater circumduction (p = 0.04), HH (p < 0.001), and SLA (p < 0.001) compared to controls, with no effect of speed (p = 0.91, p = 0.53, p = 0.72, respectively). We found speed-dependent increases in TLA in both groups (p < 0.001), however, controls had greater TLA compared to people post-stroke (p < 0.001). We also found speed-dependent increases in pKF in both groups (p < 0.001), but controls exhibited larger increases in pKF magnitudes for a given change in speed than people post-stroke (Speed x Group interaction: p < 0.001).

Walking at faster speeds increased TLA and pKF without exaggerating compensatory deviations in people post-stroke. However, between-group differences in pKF became larger at faster speeds with people post-stroke exhibiting smaller pKF than controls. This suggests that pKF deficits post-stroke may not be adequately addressed by fast walking, and may need to be specifically targeted in the context of high-intensity gait training to address both activity limitations and kinematic impairments.

Specialized roles of each hand during a mechanically coupled bilateral task in patients with mild hemiparesis and age-matched typical adults

Shanie Jayasinghe ¹ , Candice Maenza ^1,2 , David Good ¹ , Robert Sainburg ^1,2

¹Pennsylvania State University College of Medicine, Hershey, USA. ²Pennsylvania State University, State College, USA

Most upper extremity mediated activities of daily living (ADL) require coordinated interaction between the two arms, such as when donning and doffing clothing, buttoning a shirt, cutting a slice of bread, or opening a jar. Nevertheless, typical standard of care in rehabilitation for stroke survivors focuses on remediation of unilateral movements, while bilateral actions are relegated to compensatory therapy focused on carrying out ADL. Previous research on bilateral upper extremity training in stroke survivors has shown small improvements in motor coordination when compared with unilateral training alone, but these differences do not generalize to substantial improvements in functional independence. These approaches have typically used movements that do not reflect the distribution of functions that occur in everyday tasks, such as stabilizing a jar with one hand, while turning the lid with the other, nor have they addressed the natural type of mechanical coupling that occurs between the arms in such tasks. We now present a virtual reality-based task that incorporates mechanical coupling, requiring that one hand stabilize against a spring load that varies with the displacement of the other arm. We examine interlimb differences in these two aspects of control in age-matched typical adults (n=7) and chronic stroke survivors with mild contralesional impairment (7 right hemisphere damage [RHD], 8 left hemisphere damage [LHD]). All participants were right handed, as confirmed by Edinburgh Inventory. The hands were connected by a spring, and participants had to maintain the position of one hand while using the other hand to perform a slicing movement. The task consisted of a rapid out-and-back movement of one hand, with the explicit requirement of reversing the movement in a target, while maintaining a stable position with the other hand. Our two primary measures were the positional stability of the stabilizing hand and the target accuracy (reversal position) of the reaching hand. Consistent with previous research, we found that the left hand showed greater error in the out-and-back movement, while the right hand showed greater error in stabilizing, regardless of group. We also found a main effect of group for trajectory control measures, but no interaction between group and hand. These findings indicate that mild paresis results in some deficits in trajectory control while mostly preserving the specialized roles of each hand for the tested aspects of bilateral control.

Capturing the dynamics of stroke gait impairments towards developing a generative model of gait

Taniel Winner ^1,2 , Trisha Kesar ² , Gordon Berman ² , Lena Ting ^1,2

¹Georgia Institute of Technology, Atlanta, USA. ²Emory University, Atlanta, USA

Understanding the dynamics of human gait is essential to designing tailored rehabilitative therapies for individuals with gait dysfunction. Moreover, to augment or modify gait dynamics using technological devices such as exoskeletons, the ability to build predictive dynamical models of gait is essential. Currently, however, clinical researchers typically collect full continuous gait kinematic and kinetic data but only use discrete summary variables (e.g. ankle angle at paretic toe-off, peak swing knee flexion, and peak anterior ground reaction forces) [1] to characterize individual gaits, ignoring the continuous biomechanical dynamics that generate them. To test whether studying the full gait data could lead to better rehabilitative strategies, we first asked whether able-bodied and stroke gait could be better distinguished using discrete summary variables versus continuous gait kinematics. We then tested whether generalized gait dynamics (gait signatures) extracted from a Recurrent Neural Network (RNN) generative model of gait kinematics could provide for better discrimination across the groups and between individuals. We took motion capture data from 5 able-bodied and 8 stroke survivors and formulated the three different data types: 1) 26 discrete summary variables (13 averaged variables for each limb including ankle moment peak, ankle power peak, and double-support duration), 2) Six continuous joint angles (left and right limb hip, knee, and ankle flexion angles) and 3) gait signatures. To develop the gait signatures we trained an RNN model of gait kinematics and then extracted the model’s internal activations, applied principal component analysis to the internal activations, and phase-averaged the 1st six principal component magnitudes. We constructed Support Vector Machine (SVM) classifiers for the three data types in both gait group (able-bodied or stroke survivor) and for individual classification. We found that the gait signatures characterized gait the best. The SVM gait group classification accuracy of the gait signatures and discrete variables was averaged 100% ± 0.0% compared to raw continuous kinematics (93.9 ± 4.1%). The SVM individual classification accuracy of the gait signatures was averaged 99.4 ± 2.3% compared to that of the discrete variables (98.0 ± 4.0%) and raw continuous kinematics (83.8 ± 8.7%). It is noteworthy that these gait signatures, which are derived from only the continuous kinematic data, are more robust than using 26 discrete variables (which use variables like forces that are not kinematics and are not as easily measured). In the future, we will test different RNN architectures towards developing a generative model capable of predicting at least one gait cycle of gait kinematics. Building from these dynamical characterizations, it may be possible to identify the appropriate treatment for stroke survivors and build generative models of gait function, allowing us to generate individual-specific rehabilitative therapies.

Gut microbial dysbiosis is correlated with infarct size, edema size, and cerebral blood flow in aged rats following stroke

Tyler Hammond ¹ , Sarah Messmer ¹ , Jacque Frank ¹ , Doug Lukins ¹ , Rita Colwell ² , Ai-Ling Lin ¹ , Keith Pennypacker ¹

¹University of Kentucky, Lexington, KY, USA. ²CosmosID, Rockville, MD, USA

Accumulating evidence suggests that gut microbes modulate brain plasticity via the bidirectional gut-brain axis and may play a role in stroke rehabilitation. A severely imbalanced microbial community, or dysbiosis, has been shown to occur following stroke, causing a systemic flood of neuro- and immunomodulatory substances that can impact neuroinflammation as commensal bacteria invade the bloodstream and as intestinal lymphocytes migrate from gut-associated lymphoid tissue to the brain. Currently, no studies have been performed on aged rats to analyze longitudinal changes in the microbiome over the first month course of stroke rehabilitation and whether these longitudinal changes correlate with imaging markers of stroke including infarct size, edema size, and cerebral blood flow, making it difficult to confirm whether the microbiome could be a therapeutic target in stroke rehabilitation.

39 aged male and female rats underwent a permanent middle cerebral artery occlusion or a 5-hour temporary middle cerebral artery occlusion. Fecal pellets were collected before surgery and 3-days post-surgery. The rats underwent MRI for structural and cerebral blood flow (CBF) imaging at 72 hours. Data points were collected at 30 days for 4 of the rats. The microbiome was analyzed from the stool samples using whole genome metagenomic sequencing. Diversity and relative abundance of the bacterial communities were measured and correlated to stroke clinical and imaging outcomes.

Principal component analysis revealed that rat microbiomes seem to cluster together at baseline, but that most of the microbiomes diverge in the 3-day post stroke samples. The samples from 30-day post stroke cluster with the baseline. Species richness is significantly increased following stroke, rising from 52.5 to 58.7. The species evenness did not significantly increase in females but did in males (from 0.802 to 0.856), whose baseline evenness was higher than females 0.802 vs 0.688). Phyla changes occurred in five of the six major phyla that regularly occur in the gut, with increases in proteobacteria (3.18) and bacteroidetes (3.93), and decreases in firmicutes (0.72), verrucomicrobia (0.55), actinobacteria (0.58). Machine learning analysis revealed 13 species which were important predictors in distinguishing stroke from baseline samples. Phyla changes were associated with infarct and edema size, but not CBF. Species changes were associated with infarct size, edema size, and CBF.

We found that microbial communities are disrupted in an aged rat population following stroke, showing significantly different beta diversity, increased alpha diversity, and changes in the relative abundance of 5 of the 6 major phyla found in the gut. Changes in thirteen bacterial species were highly associated with stroke and changes in these species were also associated with increased infarct and edema size and decreased CBF. Information from this preparatory study will lay the foundation for developing stroke therapeutics that aid in stroke rehabilitation.

Intraparenchymal hemorrhage in a patient on chronic anticoagulation with warfarin for venous thromboembolism: challenges in neurorehabilitation- where do we go from here?

Viswanath Aluru, Stephen Moran, Nidhi Purohit, Gollamudi Reddy

Ochsner Clinic Foundation, New Orleans, USA

Introduction: 68 year old female with a past medical history significant for bilateral lower extremity venous thromboembolism and PE on chronic anticoagulation with warfarin presented to ED with weakness and dysarthria. INR was supratherapeutic. Imaging in the ED was evident for an intraparenchymal hemorrhage with a midline shift. INR was rapidly corrected. Patient was admitted to neuro intensive unit and was closely monitored with serial imaging. A neurosurgical intervention was not needed at that time. She was medically stable but left with residual weakness, dysphagia, dysarthria and cognitive deficits.

Results: Upon admission to the neurorehabilitation unit, she was able to participate in intensive therapy. She continued to present with weakness and cognitive deficits. A repeat brain imaging with non contrast MRI was evident for a stable hemorrhage. Warfarin was held since the time she presented to the ED. She was not on any antiplatelet agent. Non-pharmacological measures for VTE were administered.

Discussion: Intraparenchymal hemorrhage on chronic warfarin therapy is a known challenging complication. Resuming anticoagulation after recovery lies in a balancing act between preventing future thrombotic events versus recurrent hemorrhage which would be devastating to the patient. There is no clear consensus on when to restart anticoagulation or if at all safe to restart promptly in all patients especially after a major bleed.

Conclusion: There is a need to develop strategies to effectively stratify patients to safely resume anticoagulation after intra cerebral hemorrhage. While waiting to resume anticoagulation, there is also a need to develop evidence based surveillance strategies to avoid major complications. More research is needed to prevent anticoagulation induced complications and to elucidate underlying etiologies in patients in whom it might have caused intracranial hemorrhage.

The Dose-Response Effectiveness of Active Music Therapy for Upper Extremity Stroke Rehabilitation: A Systematic Review and Meta-Analysis Study

Abbey Tomlin ¹ , Sierra Archer ¹ , Gordon Warren ² , Yi-An Chen ¹

¹Department of Occupational Therapy, Georgia State University, Atlanta, GA, USA. ²Department of Physical Therapy, Georgia State University, Atlanta, GA, USA

Background: Previous studies have found that active music therapy (AMT) can improve upper extremity (UE) function in people with stroke. Actively playing a musical instrument is highly motivating compared to rote exercises due to the engaging music features. When playing instruments, such as a piano, the affected UE can be naturally involved in the practice, which leads to functional improvements. To investigate the effects and the optimal dosage of AMT in improving UE function in stroke, we conducted a systematic review and meta-analysis.

Methods: Our systematic review included searches in PubMed, Cochrane, MEDLINE, Embase, CINAHL, and ProQuest, with three key category terms: stroke, music-based therapy, and upper extremity. The review process was conducted by two individuals separately, with a third person involved if an agreement was not achieved. For the meta-analysis, data were extracted from the commonly-used assessments for UE function and finger dexterity in the field, including Action Research Arm Test and Nine Hole Pegboard Test. The intervention dosage in total minutes was also extracted from each article.

Results: Twenty-one articles were selected from the systematic review. Thirteen studies remained in the meta-analysis after excluding duplicate or missing data set (n=2), studies without the common assessments (n=4), and single-subject studies (n=2). Two meta-analyses were conducted: one included studies with a stroke control group (n=5) and one without (n=8). While the effect sizes of most selected studies were not significant, both meta-analyses showed a significant overall effect of AMT on UE function post-intervention (p=0.021 and p < 0.000). This indicated the potential benefit of AMT in stroke. There was no heterogeneity in either meta-analysis, suggesting a low variability among studies. Further meta-regression also demonstrated no dosage-response effect (p=0.663 and p=0.231). Greater treatment time did not associate with better motor outcomes.

Conclusion/Discussion: Our findings suggest that AMT has the potential to UE motor function in stroke survivors, compared to control treatments and pre-intervention performance. However, with the small number of articles and mostly non-randomized controlled trials (n=10), more high-quality research is needed to conclude a solid suggestion. In addition, the majority of the selected articles included AMT for a total of 5-10 hours (n=10). The low variability did not provide enough information for understanding the optimal dosage of AMT. Music-based therapy is a developing but promising intervention. Future research should continue focusing on this area to help recovery for stroke survivors.

Evaluating the ability of START, a telehealth-delivered therapy, to enhance speech for individuals with moderate-to-severe post-stroke aphasia and apraxia

Zoe Swann, Claire Honeycutt

Arizona State University, Tempe, AZ, USA

The ability to make functionally significant changes in individuals with severe post-stroke speech loss (aphasia/apraxia) remains a key challenge for the rehabilitation community. Here, we aim to evaluate the efficacy of Startle Adjuvant Rehabilitation Therapy (START), a tele-enabled, low-cost treatment, to improve quality of life and functional speech in individuals with severe-to-moderate stroke. START is the exposure to startling acoustic stimuli during practice of motor tasks (speech, upper extremity) in individuals with stroke. We have previously shown that START increases the intensity of practice in severely-impaired post-stroke reaching, with START eliciting muscle activity 2-3 times higher than maximum voluntary contraction, and 350 ms faster than non-startled reaching for individuals with severe to moderate stroke. Moreover, voluntary reaching distance, onset, and final accuracy increased after a session of START, suggesting a potential rehabilitative effect. However, START has not been evaluated during impaired speech. The objective of this study was to determine the therapeutic effects of startle on moderate to severe post-stroke aphasia and apraxia of speech.

Individuals with post-stroke aphasia and/or apraxia were trained with (START; N=17) or without (CONTROL; N=16) a startling acoustic stimulus. Training consisted of 3 days of remotely-delivered high-frequency auditory-repetition of words. Clinical measures of aphasia (WAB-R), apraxia of speech (ABA-2), and subject-reported quality of life (SIS) were evaluated before and after training, as well as 1 month post.

Our data demonstrate improvements in motor, cognitive, and quality of life domains after just 3 sessions. Aphasia Quotient (AQ) more than doubled under START (7.5%) compared to Control (2.7%). AQ was bolstered by increased Comprehension (START: 7.4%; Control: 3.4%), Repetition (START: 15.6%; Control: 4.91%), and Naming & Word Finding (START: 17.4%; Control: -7.0%). Training with START also doubled Apraxia measures of accurate multi-syllable repetitions in the Diadochokinetic Rate subtest (START: 38.5%; Control: 17.7%) and reduced in errors in words of increasing length (START: -36.6%; Control: -17.7%). Changes in Aphasia & Apraxia measures resulted in striking subject-reported increases in quality of life, in both Communication (START: 31.5%; Control: 13.2%) and Total Scores (START: 12.8%; Control: 0.7%).

These results suggest START has the most benefit for individuals with severe post-stroke impairments in both cognitive and motor speech domains. This fills an important gap in aphasia care, as many speech therapies remain ineffective and financially inaccessible for patients with severe deficits. Here, START is remotely-delivered and has the potential to be effective and accessible for those that have poor access to quality care. We are the first to evaluate START as a therapy and first to evaluate START in stroke. More quantitative studies are required to understand how START influences the neural and acoustic mechanisms of speech to generate these changes.