Methodological Recommendations for Studies on the Daily Life Implementation of Implantable Communication-Brain–Computer Interfaces for Individuals With Locked-in Syndrome

Abstract

Implantable brain–computer interfaces (BCIs) promise to be a viable means to restore communication in individuals with locked-in syndrome (LIS). In 2016, we presented the world-first fully implantable BCI system that uses subdural electrocorticography electrodes to record brain signals and a subcutaneous amplifier to transmit the signals to the outside world, and that enabled an individual with LIS to communicate via a tablet computer by selecting icons in spelling software. For future clinical implementation of implantable communication-BCIs, however, much work is still needed, for example, to validate these systems in daily life settings with more participants, and to improve the speed of communication. We believe the design and execution of future studies on these and other topics may benefit from the experience we have gained. Therefore, based on relevant literature and our own experiences, we here provide an overview of procedures, as well as recommendations, for recruitment, screening, inclusion, imaging, hospital admission, implantation, training, and support of participants with LIS, for studies on daily life implementation of implantable communication-BCIs. With this article, we not only aim to inform the BCI community about important topics of concern, but also hope to contribute to improved methodological standardization of implantable BCI research.

Introduction

Individuals with severe motor and communication impairment may benefit from a Brain–Computer Interface ¹ (BCI; see Leuthardt et al ² and Antonietti et al ³ for glossaries), as these devices may offer a muscle-independent communication-channel. Despite significant progress in research and development of implantable BCIs, where recording electrodes are implanted on or in the cortex,^4,5 efforts to validate this technology as a communication tool in the daily life of people with severe motor impairment remain scarce. In 2016, we described the first case of independent home use of a fully implantable communication-BCI by an individual with late-stage amyotrophic lateral sclerosis (ALS).^6,7 This electrocorticography (ECoG)-based BCI, called Utrecht NeuroProsthesis (UNP), enables users to generate “brain-clicks” for the control of assistive communication software. Another example of a fully implantable communication-BCI for settings of daily living is the Stentrode^®,^8,9 which relies on endovascular electrodes placed in the superior sagittal sinus. In addition, in the Braingate and similar trials, participants have used implanted, intracortical, electrodes (with external amplifiers) to accomplish communication,^{10 -15} but to the authors’ knowledge, independent home use for communication in settings of daily living has not been demonstrated with this type of BCI systems.

For implantable BCIs to eventually become clinically available for people with severe motor impairment, their usability as a communication tool in daily life needs to be assessed with more end-users. Besides improving our understanding of user characteristics and environmental circumstances that are supportive or prohibitive for BCI functioning, we need to validate advanced systems that provide more degrees of freedom and faster communication. However, conducting research on home use of implantable communication-BCIs with people with severe motor impairment is highly complex, which likely is an important cause for the very limited number of clinical trials on this topic. We believe that rigorous validation of implantable communication-BCIs is served by an increase in the number of research groups undertaking the effort, and by methodological standardization of these efforts. To this purpose, we here describe a set of procedures and recommendations for screening, inclusion, and implantation of people with severe motor impairment that may benefit research groups that for the first time consider to translate their fundamental BCI research to validation studies on implanted BCIs for communication (or other applications). Importantly, whereas this article advocates an approach that is generalizable, some considerations inevitably relate to the specific hardware used. For details about the UNP study, on which our experiences are based, and in which 6 individuals were enrolled, of whom 3 received a fully implantable ECoG-based BCI system, we refer to the original publication. ⁶

Target Population of Implanted Communication-BCIs

The primary target population for (implanted) communication-BCIs are people who experience, or face experiencing, locked-in syndrome (LIS), which is characterized by an inability to move and speak in the presence of intact cognition. ¹⁶ LIS is typically linked to events that cause pontine damage (eg, stroke or injury), but also demyelinating and neurogenerative diseases (eg, ALS, central pontine myelinolysis) may lead to this condition.^{17 -21} A communication-BCI may provide a solution for the substantial portion of people with LIS for whom conventional, muscle-controlled, assistive communication technology, such as eye-gaze devices, falls short,^{22 -25} because of difficulties keeping the head still, ²³ pupil dilation ²⁶ or other side effects of medication, progressive oculomotor impairment, ²⁴ and/or eye gaze fatigue. ²³ The interindividual variability in user preference regarding the various practical and functional characteristics of communication-BCIs that are based on non-invasive and implanted electrodes,^{27 -29} underline the importance of the continued development and validation of both approaches. In this article, we focus on implanted communication-BCIs for people with LIS.

How to Reach and Inform Candidate Participants?

As LIS is defined according to functional criteria ¹⁶ and has a diverse etiology,^{17 -21} it is difficult to reach this target group and inform them about BCI studies of interest. Additional complicating factors are that (in Europe) most people with LIS live at home^{21,30 -32} and no longer receive specialist treatment. Recruitment of potential study participants with LIS therefore benefits from involving multiple channels, including general media, general practitioners, specialists, patient organizations, and home ventilation teams. Upon relevant permissions, screening (neurology) databases for people with relevant diagnoses may be helpful as well.

For most research, informing potential participants about study details is accomplished via a participant information brochure and a conversation with the researcher. Given the communication impairment of implant candidates, the large amount of information that needs to be processed by them, and the significant impact participation will have on their life and that of their close relatives, informing candidates with LIS about participating in a study on implanted BCIs, however, should be undertaken as an iterative process, where information is provided in stages, where the candidate is asked, in each stage, if they are still interested, and where all information and communication is titrated to the candidate’s limited communicative abilities. Besides an information brochure, an information video and website can be offered. Since traveling to the research institute may be complicated for people with LIS, the research team may want to conduct one or more home visits during the information process. This also allows for ample time to discuss the details of the study, to answer any questions candidates and their family/caretakers may have and, importantly, for the research team to become familiar with the communication abilities and needs of the candidate and the setting the BCI will be used in.

What Eligibility Criteria Should Be Considered and How Should They Be Assessed?

Several eligibility criteria must be considered for studies on implantable communication-BCIs in people with LIS:

1. Reliable communication channel: Although implanted communication-BCIs are relevant for those who have lost all muscle-based communication, ¹⁵ there are important advantages to enrolling participants with progressive diseases such as ALS in research before this stage is reached, including the opportunity to obtain first-person consent, an assessment of cognitive functioning in advance of the surgery (see below) and feedback on, for example, user satisfaction, fatigue, and motivation (see below) throughout the study. Therefore, we advise that candidates should have at least one reliable channel to answer closed (yes/no) questions, for example, by using eye blinks. This communication channel is essential for candidates to give informed consent, to partake in (neuro)psychological evaluation, to communicate about care needs during hospital admission, and to give feedback to the research team during BCI training. Without such communication channel these interactions become virtually impossible.

   By asking the potential candidate to answer, with their preferred communication channel, a set of trivial yes/no questions to which both the candidate and the researcher know the answer (eg, “Is your name John?”), one can assess the ability of the candidate to hear, understand, and answer these types of questions, and become familiarized with the candidate-specific yes/no signals.^33,34 If there is doubt about the interpretability of the answers, this procedure can be repeated in a single-blind fashion, with a new set of questions to which the candidate, but not the researcher, knows the answers. A caregiver should be involved in designing these questions and in verification of the correctness of the answers. Further screening and informed consent can only proceed if the responses of a candidate can be interpreted unambiguously.

   During screening, it is advisable to discuss current and previous (attempts to) use conventional assistive communication technology and to refer candidates to an assistive technology expert in case they seem inadequately informed about conventional options.

2. Stable neurological condition: Since research on home use of communication-BCIs likely involves long-term study participation, the neurological condition of LIS participants preferably is relatively stable. For conditions such as brainstem stroke or spinal cord injury, ⁹ one may want to await the extent of recovery of motor- and communication-impairment before deciding to implant a BCI. Spontaneous functional recovery after stroke, for example, can render the BCI implant obsolete, and typically occurs in the first weeks to months, with the time course being related to the severity of the stroke.^35,36 To be on the safe side, a period of 1 year ⁹ can be taken to determine if someone remains functionally impaired at a level that justifies BCI implantation. In the case of ALS, progression rate varies across individuals. ³⁷ Fast disease progression may cause participants to lose their muscle-based communication channel during the study, which could slow down research progress (although such a situation does not necessarily make it impossible to accomplish BCI-based communication ¹⁵ ). In addition, there are indications for a relationship between ALS progression rate and cognitive decline.^{38 -41} Researchers may consider consulting the candidate’s neurologist about their disease progression rate and expected survival ³⁷ and exclude individuals with an unfavorable prognosis.

3. Minimum level of cognitive functioning: Some causes of LIS may also cause cognitive impairment. Up to 15% of people with ALS develop frontotemporal dementia ⁴² and also people with brainstem stroke may show a decline in cognitive functions (see for review, D’aes and Mariën ⁴³ ). LIS participants should at least be mentally competent, which is a legal requirement for informed consent, and thereby able to understand and evaluate the benefits and risks of study participation. A certain minimum level of cognitive functioning is also essential for understanding and following task-instructions, and for BCI training. Cognitive abilities can be assessed with a neuropsychological evaluation. Given the communication problems of people with LIS, a dedicated test battery should be assembled that relies on multiple choice or yes/no options, to evaluate several key aspects of cognition (eg, Raven’s Advanced Progressive Matrices for general intelligence, Peabody Picture Vocabulary Test for speech comprehension, and Visual Association Test with multiple choice options for memory) within a reasonable amount of time. Assembling this test battery and running it with a participant preferably involves a certified neuropsychologist. Comprehension of specific study-related topics can be evaluated by asking the candidate to answer a list of closed questions on these matters. ³³

4. Low depression score: Research indicates that quality of life is often perceived as relatively high by people with LIS due to ALS^44,45 or brainstem stroke, ⁴⁶ and is not necessarily related to the level of physical impairment.^{46 -48} Yet, people with ALS and brainstem stroke can go through phases of depression.^{49 -52} Severe depression may negatively affect the capacity to make decisions about treatment or research participation ⁵³ and is likely to influence motivation in the course of the study. We therefore consider it important to screen for depression and suggest to use the ALS Depression Inventory (ADI-12 ⁵⁴ ) for such assessment and to exclude candidates with a score of 30 or higher (ie, severe depression).

5. Surgical clearance: As with all surgeries, candidates should receive surgical clearance before proceeding to the operating room, to minimize surgical risk. After permission of the candidate, the relevant medical history should be made available to the study physicians (eg, neurosurgeon and anesthesiologist) to allow for risk assessment. Important considerations for BCI implant candidates include a medical contraindication to stop anti-coagulant medications during surgery, brain damage (eg, cortical damage or anomaly at the electrode target location, or a pre-existing condition that increases surgical complexity, such as previous neurosurgery, stroke, or intracranial infection), history of brain tumor, or other conditions that necessitate regular follow-up magnetic resonance imaging (MRI) scans (not possible with an MRI incompatible BCI implant), cancer, malnutrition, active infection or inflammation (including decubitus ulcer), significant cardiovascular, metabolic, or renal impairments, allergies and the respiratory situation. The last item is of special concern for candidates with neuromuscular disease, since respiratory insufficiency, which is common in this group, may be associated with increased surgical risk ⁵⁵ and a potential risk of not being able to wean from intraoperative mechanical ventilation.

6. MRI compatibility: A BCI implant candidate should be able to undergo a (functional) magnetic resonance imaging ([f]MRI) scan to evaluate the suitability of the candidate for BCI implantation, presurgically localize the electrode target area, and assess the presence of brain damage or atrophy. General MRI safety issues should be met and the participant must be able to lie in a supine position for the duration of the scan without problems related to severe discomfort, dysphagia, or respiration.

7. Vision and oculomotor function: As many BCI training applications and user interfaces rely on visual information, significant vision loss can represent an obstacle for BCI training and home use. However, vision loss is not necessarily an exclusion criterion, since auditory user interfaces can be implemented as well. ⁵⁶ When a candidate experiences significant oculomotor impairment, such as in late-stage ALS, ⁵⁷ visual information may need to be presented in the center of the visual field. Involving an ophthalmologist in the screening stage may contribute to an adequate assessment of these factors and to an optimal preparation of BCI user interfaces.

8. Favorable living situation: For studies that aim to validate the use of BCIs in settings of daily living, family members and caregivers are inherently involved in the research and at-home tests. Their practical and emotional support will be essential for study success and may contribute to keeping the participant motivated when progress is slow. To facilitate this, resident family members and/or a primary caregiver should be actively involved in the information process, among others to discuss mutual expectations. Researchers may also consider providing a dedicated information letter to these individuals. In addition, given the typically frequent research visits, participants preferably live within a reasonable distance from the research institute.

Since some of the abovementioned criteria (eg, cognitive functioning) cannot be evaluated without significant involvement of the participant, part of the screening will have to take place after informed consent. Candidates should therefore be informed about the possibility that they may be excluded based on the post-informed consent screening steps.

How to Obtain Informed Consent From People With Severe Communication Impairment?

As people with LIS are unable to sign an informed consent form, obtaining and registering informed consent requires special attention. The participant’s ability to understand and evaluate information, and to communicate consent in an interpretable manner, should be assessed during screening (see above). For archiving and registration purposes, we advise to record the formal moment of asking informed consent on video. ³³ We also recommend that an independent observer attends the procedure, and verifies the proper execution thereof and the responses given by the candidate. ⁵⁸ The procedure itself may include the following steps: (1) summary of study-related information; (2) registration of the interpretability of the candidate’s yes/no responses and understanding of the study procedures, by asking the candidate to once more answer the 2 lists of closed questions described in sections “Reliable communication channel” and “Minimum level of cognitive functioning”; (3) if the questionnaires are answered accurately: asking consent to participate in the study. We propose to ask this question 3 times and to require 3 clear, affirmative answers as a definitive expression of consent. The legal representative of the candidate then signs the consent form on behalf of the candidate, followed by the independent observer and researcher. When the formal moment of informed consent is well prepared for with an iterative information process, surprises are unlikely to occur. However, because of fatigue, excitement, or otherwise, communication of the candidate may be less clear than before. In that case, one may consider interrupting and rescheduling the procedure.

How to Decide Where to Place the Electrodes?

The Value of fMRI

The chance of obtaining adequate BCI control is likely improved if electrodes tap into a brain area that shows clear task-related signal changes.^59,60 fMRI has been used for non-invasive presurgical assessment of BCI electrode target locations,^6,9,61,62 since the Blood Oxygen Level Dependent fMRI signal provides high spatial detail (ie, in the order of millimeters) and shows a close spatial correspondence with important neuro-electrical signal features for BCI.^{63 -65} A clear task-related fMRI activation pattern can also be confirmative about the participant’s capacity to understand instructions and execute the tasks that are necessary to accomplish BCI control, and about their ability to sustain attention for the duration of the fMRI scan. Indeed, these factors are known to contribute significantly to the success of fMRI investigations with pediatric neurosurgical populations.^66,67 As such, fMRI may contribute to eligibility screening: implantation surgery is then scheduled only for those participants who show clear fMRI activity and are therefore likely to accomplish BCI control. As a rule of thumb, the fMRI activation can be used for surgery if t-values exceed at least a level of 4 (given only the sensorimotor cortex is of interest), displays clear and clustered activity in the target area and no motion artifacts.

When designing the fMRI scan protocol, given the complexity of scanning individuals with LIS, we recommend to include block-design tasks (with at most 3 conditions) that are easy to understand and conduct and that have proven useful in clinical settings (see for instance Jansma et al ⁶⁸ ). In addition, one may add tasks to assess the spatial discriminability of multiple motor or speech-related conditions to the fMRI scan session. For all tasks, clear instructions and practice may contribute to a usable result.

Besides functional scans, the scan protocol preferably contains a structural T1-weighted scan and an angiographic scan. The T1 scan can be used to visualize fMRI activity patterns in relation to the individual’s sulcal/gyral organization, and to assess the presence of brain damage or atrophy. The angiographic scan will inform the researcher on the presence of large blood vessels in the electrode target area, which may be helpful in deciding about the target electrode location or orientation, since blood vessels under ECoG-electrodes attenuate the recorded signal. ⁶⁹

Taken together, structural and functional imaging can play an important role in deciding where to place electrodes, and in avoiding the risk of brain surgery for people who are unlikely to reach adequate BCI control.

Acquiring (f)MRI Data From People With LIS

A first factor to consider when performing (f)MRI scans with people with severe motor impairment or LIS is respiratory support. If a participant receives artificial home ventilation (non-invasively or via tracheostomy), the home ventilation machine needs to be replaced temporarily by an MRI-compatible machine. Transferring between machines needs to be performed by clinicians with relevant expertise (eg, anesthesiologist or pulmonologist), who should be also responsible for vital function monitoring during the scan and for providing support (eg, airway suctioning) when needed. Ideally, clearing of the airways using a cough-assist or suctioning is performed immediately before entering the scanner (ie, after transfer of the participant onto the scanner bed) to minimize chances of discomfort and interruptions of the scanning procedure.

A second factor is the supine position in the scanner. This position may cause discomfort around the tracheostomy, which may be (partly) resolved with a thin pillow. In addition, attention must be paid to excessive salivation and swallowing problems, although aspiration is not expected to happen for participants who are ventilated with a cuffed canula.

Third, communication with the participant should be guaranteed at all times and requires special attention for those who are unable to speak or press the MRI-alarm button. During the transfers and the scan itself, the use of assistive communication technology will be limited or impossible, and communication needs to rely on closed questions and a letter card (card with matrix of letters; the participant blinks or moves the eyes to select target letters pointed at in sequence by a communication partner), limiting the options of the participant to self-initiate communication or ask for help. Therefore, a team member should be responsible for communication, constantly attending the participant’s communication channel and regularly asking about wellbeing. During the scan, the communication channel can be monitored with an MRI-compatible camera. Using an agreed-upon signal (eg, continuous blinking), the participant can request attention at any time. A dedicated question tree that focuses on vital issues first helps to quickly find out what kind of support is needed.

Localization of Target Areas for Placement of Electrodes

For implantation of small electrode arrays, spatial precision is crucial. The results of fMRI data analysis can be used to identify foci of task-related activity, which can be visualized on cortical surface renderings generated from the T1-weighted scan. Then, virtualized electrodes can be projected onto the renderings such that relevant foci are covered. Based on the desired position of the electrodes, a craniotomy or burr-holes can be planned.

To transfer the positions of the planned craniotomy/burr-holes and electrode target areas from MRI space to that of the participant’s head at the operating table, neuronavigation procedures can be used. To that purpose, a neuronavigation MRI or computed tomography (CT) scan should be acquired shortly before surgery. Because of the complexities of conducting MRI scans with people with LIS and respiratory support, a neuronavigation CT scan with adhesive fiducials is preferred. The coordinates of the planned craniotomy/burr-holes and electrode targets are registered to the acquired neuronavigation scan, the results of which can be uploaded to the neuronavigation device in the operating room.

What Factors to Take Into Account When Scheduling Hospital Admission?

When a participant successfully passed all screening steps, hospital admission and surgery can be scheduled. Depending on the participant’s respiratory situation and on local guidelines, participants may have to be admitted to medium or intensive care facilities. Before admission, the respective department should be thoroughly informed about the situation of elective brain surgery for implantation of an investigational medical device, and the participant’s special needs and limited communication abilities. An important topic to address is the possibility of connecting a participant’s home ventilation system to the hospital’s nurse call system, so that hospital staff is called automatically in case of ventilator-related problems. In addition, drawing attention of, and communication with, hospital staff should be considered. When a participant has some remaining muscle activity, it should be inventoried whether the ward has special alarm buttons available that could be activated with this activity. Since it will not always be possible to use assistive communication technology, a simple letter card at the bedside will be helpful. Third, the necessity and possibility of rooming-in of a spouse or nurse of the home care team should be discussed, as well as the option for this individual to conduct (some of) the standard care they perform at home. In our experience, having a relative or caretaker available 24/7 during the hospital stay contributes greatly to the mental and physical comfort of the participant, and facilitates communication and understanding between hospital staff and the participant. Finally, a researcher should stay in close contact with the hospital staff and the participant and their relatives, to keep the research team up to date about the participant’s condition and to make sure that all parties are informed about the scheduled research procedures.

How to Ensure Electrodes are Placed Precisely on the Target Locations?

BCI implantation is a complex procedure. Apart from the surgical technical aspects, which may be more or less familiar to the surgeon, the research setting in which the surgery is conducted requires specific attention. Before implanting the first participant, surgeons with little experience with implanting electrodes in/on the brain may consider attending implantation surgeries by experienced colleagues, and the surgical procedure may be tested on a 3D printed model of a skull and brain. Such a dry run may also contribute to defining the roles and responsibilities of the people present at the operating room. Responsibility for all medical and surgical aspects is with the neurosurgeon and their team. The researcher is responsible for advising where the electrodes should be placed, to confirm the electrodes, leads and amplifier are correctly connected (see below), and for registering the location of each implanted electrode array. For the actual implantation surgery, the to-be-implanted parts will typically be provided ready-for-use by the manufacturer (sterile and packaged appropriately). It may be helpful to have an implanted device manufacturer representative present or on call for technical support.

It must be ensured that electrode placement does not diverge from the fMRI-based surgical plan (unless for medical reasons). In case burr-holes are used, precise neuronavigation-guided drilling thereof is essential. In case of a craniotomy, neuronavigation may or may not play a determining role: subdural or intracortical placement has the advantage of having the cortical surface visible but in case of epidural placement the sulcal/gyral/vascular pattern is invisible, requiring neuronavigation.

After electrode placement, impedance measurements may be used to assess lead integrity and proper electrode contact with brain tissue. In addition, the electrodes may be temporarily connected to an external (clinical) data acquisition system, to evaluate signal quality and to signal any lead defect before closing the skin. For this purpose, it may be considered to collect somatosensory evoked potentials, or to wake up the participant and ask them to conduct a functional mapping task, according to standard awake neurosurgical procedures. Finally, in case an implantable amplifier is used, proper functioning thereof should be confirmed using manufacturer guidelines before closing the skin. It is recommended to have spare components available in the OR to replace malfunctioning components.

The final electrode position relative to the cortical surface can be established using a postsurgical CT scan with high contrast for electrodes and skull.^70,71 The CT scan can be aligned to the pre-operative T1-weighted MRI scan and the 3D-coordinates of the electrodes can be computed and visualized on, for example, a pial surface rendering. The CT scan will also be informative on the development of any surgical complications, such as subdural hematoma.

Pre- and postsurgical neurological evaluation can be used to assess and register any surgery-related changes in the participant’s remaining motor and somatosensory functions.

Post-Implantation Monitoring

After discharge to home, postsurgical clinical monitoring can be conducted according to clinical standard. In case of a chronic percutaneous connection, the wound will require regular cleaning and inspection. In addition, monitoring of the signal quality is important to signal any problems related to implant integrity or tissue reaction. Impedance measurements and evaluations of baseline signals and task-related modulation of the signals can be helpful for this purpose.

How to Train Participants to Use a BCI for Communication (or Other) Purposes?

BCI training is an iterative process where the participant learns to use their neural signals for computer control while BCI settings are optimized. Since there is no one-size-fits-all solution for this process, we here describe several topics of more general concern.

1. Training in settings of daily living. There are significant advantages to conducting BCI training of people with LIS during home visits, and several studies on implanted BCIs have used this approach.^9,58,72 First, many people with LIS fatigue easily. If a participant does not need to travel to a research institute, they may be able to conduct more research tasks during a session and concentration may be better. Second, since communication-BCIs eventually need to be usable in daily life, which includes many sources of distraction and environmental/electrical noise that may affect the neurophysiological recordings, testing algorithms, and user interfaces in that setting will provide crucial information for eventual real-world implementation thereof. Researchers may want to pay particular attention to acoustic contamination ⁷³ and to electromagnetic interference emitted by medical devices in the participant’s environment. When conducting BCI training at the participant’s residence, researchers enter the private environment of the participant. Adequate communication between and during research sessions, with the participant, family, and caregivers, is important for optimal training progress and a satisfactory participant experience. Studies on implanted BCIs in people with motor impairment typically schedule 1 to 3 research and training sessions per week.^{6,9,14,72,74,75} The actual frequency of research visits will, however, depend on participant availability, ability, and willingness. Session duration may depend on the training stage, but will eventually be determined by the participant, considering, for example, fatigue. Fatigue levels can be assessed at the onset and end of each session using a visual analogue scale and session durations can be adjusted accordingly.

2. Feedback. Especially the initial stages of BCI training may require participants to repeatedly conduct screening tasks (eg, attempted movements alternated with rest), to determine which electrodes show strong and consistent signal changes. It is advisable to also implement simple BCI-control tasks, in which the participant receives feedback about their brain signal, in the training program from the start. ⁷⁶ In addition, an engaging training environment, with a variety of attractive (game-based) tasks, may contribute to optimal training. ⁷⁶

3. Participant-centered parameter optimization. BCI training procedures typically aim to maximize accuracy and speed. These parameters are often taken together in a metric called Information Transfer Rate. For communication-BCIs, we recommend to (also) report the number of correct characters/words per minute, which is an intuitive metric that has been used by several earlier studies on this topic.^6,9,10,77 In addition, while acknowledging the importance of these factors, we advocate a participant-centered approach to parameter optimization in studies on the daily life implementation of BCIs, as preferences in terms of the balance between speed and accuracy may vary, and since perceived usability of the BCI is also determined by other factors, such as workload or effort. ⁷⁸ These other factors require separate assessments: information about experienced effort, for example, is typically not an output of BCI feedback tasks. Regular interaction with the participant about these factors fosters optimal user satisfaction and success of the BCI.

4. Factors affecting performance. Several studies have suggested that factors such as fatigue ⁷⁹ and motivation ⁸⁰ may affect BCI performance, but we still lack a complete understanding of how participant-related and environmental factors benefit or impede BCI control. Since such understanding will be important for future clinical implementation of implantable BCIs, we advise to assess and register factors such as mood, motivation, and fatigue during every visit. This also facilitates signaling of changes in psychological wellbeing that could be relevant for the study or that may require physician referral. In addition, it is good practice to carefully log remarks and observations related to the participant’s physical and psychological wellbeing (including any changes in remaining muscle movement ability, medication use, and [recovery from] illness), as well as the setting in which measurements were made (eg, participant positioning, time of day etc.).

Data gathered in implantable BCI clinical trials is highly valuable, since this setting provides a unique opportunity to acquire longitudinal brain data recordings. These data should be acquired and stored with high quality standards for clinical data acquisition and data provenance. Protection of privacy of data gathered in daily life settings requires specific attention, because of the possibility of reconstructing private conversations based on home use data of a communication-BCI.

Technical/Consumer Support

Although it should be made clear to candidate participants before enrollment that the chance that they experience benefit from research participation is small, such benefit may nevertheless appear once a participant is capable of reliable BCI control and is provided with a system for independent home use. We recommend that the BCI (at least initially) does not replace any conventional communication technology used by the participant, but is offered as a parallel tool. Importantly, the BCI system is a medical device and must comply with Food and Drug Administration and/or European Union regulations on class III medical devices. Some countries may apply less strict rules to devices used in the context of a clinical trial than to a marketed device. Any software that is developed by the research team, and is part of the home use BCI system, needs to be considered as part of the medical device as well. Design and development of such software requires quality control at a level that is as close as possible to locally applicable requirements for software as part of a medical device. Such quality control needs to take into account that risks during home use may be different than under controlled circumstances in a laboratory. An important topic of concern in this regard is the level of access users or caregivers have (eg, adjusting parameter settings).

It is mandatory to provide written instructions for use to both users and caregivers, and, if necessary, to provide hands-on training. Typically, the manufacturer of the implant and associated hardware only provides written instructions for use for professionals. Since this information is not readily accessible to most users, we recommend to include, in the written instructions for the user, a laymen’s version of the most relevant parts of the manufacturer information. Besides information on how to use the system, the instructions for use may contain warnings on how to avoid BCI system failure.

Despite high quality design and development, detailed instructions for use and training, occasional technical support is inevitably necessary. In addition, as the clinical trial is a way to validate the BCI system, user feedback may provide guidance for further system improvements. Setting up the research with an adaptive trial design ⁸¹ facilitates an iterative process, where information collected during the study, including participant feedback and error reports, can be used to make adaptations to the study.

Importantly, the Declaration of Helsinki ⁸² states that provisions should be made “for post-trial access for all participants who still need an intervention identified as beneficial in the trial.” This means that continued access should be facilitated for those who benefit from the BCI after research participation ends. Such post-trial access should be prepared for in advance of the study, as it will require significant resources for continued support.

Discussion and Conclusion

We described important considerations for studies that test and validate implantable communication-BCI systems in settings of daily living of people with LIS. Importantly, the success of such studies will depend on a detailed preparation of each of the described stages, before the study commences (eg, when drafting the institutional review board proposal or investigational device exemption application), and during the study, when procedures are scheduled for a specific participant. In that preparation, it is crucial to involve clinicians with relevant expertise, so that standard pre-, peri-, and post-operative clinical procedures (eg, administration of antibiotics, wound care) are followed where possible and researchers and clinicians are well-aware of where, why, and how deviations from these procedures are implemented, and who is responsible in these cases. It goes without saying that communication is essential, and that drafting standard operating procedures can be helpful. In addition, it should be acknowledged that participation in this type of research represents a significant physical and mental challenge for the participant and their family. To optimize the experience of research participation, and to maximize chances of study success, the research team should, at all times, consider the perspective of the participant and their family and try to address expressed needs.

Although the recommendations described here are based on the UNP clinical trial, which employs subdural electrode strips, they are more widely applicable. For example, the procedures to define and accomplish optimal electrode positioning may also be relevant for studies that aim to test high-spatial density ECoG grids, which increase the versatility and complexity of decoding.^{77,83 -86} Also in cases where (ECoG) electrodes are implanted for BCI purposes other than communication, such as stroke rehabilitation, similar considerations and procedures may apply. Other types of electrodes, for example, those that target blood vessels (eg, Stentrode^®) ⁹ or subcortical regions (eg, NeuroPace^®), ⁸⁷ may require additional steps specific to the implant. The procedures we implemented for screening, informing, and inclusion of people with LIS, as well as for conducting (f)MRI scans and scheduling hospital admission, may be applicable for other studies targeting this population, also outside of the BCI field.

In summary, we propose a set of procedures and recommendations that may benefit future studies on the daily life implementation of implantable communication-BCIs for people with LIS. These considerations may not only serve as a guidance for future studies, but may also contribute to improved standardization thereof, enabling joint analyses and increased comparability of results, which is specifically relevant for the assessment of how the different participant-related and environmental factors affect BCI control, the understanding of which is crucial for eventual effective clinical implementation of implanted BCIs.