Abstract
Reduced toe clearance during the swing phase of gait, often referred to as foot drop, is a common cause of walking disability in clinical populations like stroke, cerebral palsy, or multiple sclerosis. Individuals with foot drop often wear an ankle-foot orthosis (AFO) to prevent excessive plantarflexion, but many commercially available AFOs overly restrict ankle mobility or make the wearer feel unstable/uncomfortable. Soft AFOs—AFOs with soft attachment points and elastic assistance—are designed to retain ankle mobility and comfort. However, their effect on gait biomechanics, as compared to traditional AFOs, is not well understood. Therefore, the objective of the current study was to perform a comprehensive biomechanical and neurophysiological comparison of soft AFOs with traditional AFOs. Sagittal plane kinematics, ground reaction forces, and lower extremity muscle activation were measured in 23 neurologically intact individuals while walking on a treadmill without assistance from an AFO (No AFO) and then with unilateral assistance from four commercially available AFOs (rigid anterior, flexible posterior leaf spring, and two soft AFOs). We found that soft AFOs allowed for greater ankle dorsiflexion velocity, plantarflexion velocity, and plantarflexion angle while retaining or increasing dorsiflexion during the swing phase. We also found that the traditional AFOs reduced propulsive ground reaction forces compared to the soft AFOs, and the rigid anterior AFO reduced plantarflexor muscle activity compared to the soft AFOs. These results highlight the differences between different commercially available AFOs and present soft AFOs as an exciting alternative to traditional AFOs when ankle mobility is desired.
Introduction
Neurological disorders such as stroke, multiple sclerosis, and cerebral palsy often result in impaired motor control that impacts gait, resulting in inefficient and/or unstable walking patterns and slower gait speeds (Booth et al., 2018; Comber et al., 2017; Langhorne et al., 2009). One particularly prevalent sequela of these conditions is reduced toe clearance during the swing phase of gait resulting from poor voluntary drive to the dorsiflexors, abnormal plantarflexor activation, and/or altered hip and knee flexion (Barrett et al., 2009; Klein et al., 2013; Knorr et al., 2011; Little et al., 2014; Sakuma et al., 2014). This reduction in toe clearance is often referred to in clinical environments as “foot drop” and is considered to contribute to post-stroke gait dysfunction, although other factors, such as impaired limb shortening, are known to affect post-stroke limb advancement (Little et al., 2014). Critically, foot drop can increase an individual's risk of tripping and falling (Barrett et al., 2009; Mao et al., 2022). As a result, devices that influence ankle dorsiflexion, especially during the swing phase of gait, are of critical interest to clinicians and patients alike.
The most common type of device for addressing foot drop is passive ankle-foot orthoses (AFOs), which provide support/assistance to ankle dorsiflexors via passive structures (e.g., composite materials, cables). Passive AFOs have been shown to improve gait speed, cadence, step length, and stride length in clinical populations (Aboutorabi et al., 2017; Choo & Chang, 2021b) and are generally more cost-effective than active devices like robotic AFOs or functional electrical stimulators. Most commercially available passive AFOs can be roughly categorized into three categories: Rigid, Flexible, and Articulated. Rigid AFOs use inflexible structures made of a composite or high stiffness plastic materials to completely immobilize the ankle. However, fully restricting ankle plantarflexion inhibits natural ankle movements, which can reduce the propulsive capabilities of the impaired limb during gait thereby decreasing gait efficiency (Totah et al., 2019). Reducing natural ankle motions can also increase injury risk by inadequately assisting balance recovery (Nevisipour & Honeycutt, 2022) or interfere with daily activities such as stair-climbing or driving (Dinkel et al., 2023). Ankle immobilization also raises concerns over promoting disuse atrophy (Appell, 1990) that are indirectly evidenced by observed reductions in electromyographic activity (Geboers et al., 2002; Lairamore et al., 2011). Furthermore, full ankle immobilization with poorly fitting AFOs can make the wearer feel unstable, uncomfortable, and lead to pressure sores (Zaino et al., 2023). Because of this, rigid AFOs are generally considered to be most appropriate for more severely impaired individuals (Delafontaine et al., 2017). Flexible AFOs, such as the posterior leaf spring (PLS), attempt to counter these shortcomings by using more compliant materials (e.g., silicone or lightweight thermoplastics) to offer greater range of motion and propulsion. However, despite the use of more compliant materials, flexible AFOs also discourage natural ankle motion and can therefore restrict propulsion as well (Desloovere et al., 2006; Ounpuu et al., 1996). Furthermore, flexible AFOs often fall short of providing comprehensive support, and are therefore only considered appropriate for individuals with moderate impairment (Choo & Chang, 2021a). Articulated AFOs prevent plantarflexion while allowing for some amount of voluntary dorsiflexion, often with a hinge (spring-loaded or free) or cable. While innovative, articulated AFOs lack robust clinical evidence, are too cumbersome for daily use, and suffer the same drawbacks as rigid AFOs due to the restriction of plantarflexion.
Recently, soft AFOs—systems composed of soft bands connected by elastic elements that assist dorsiflexion and/or plantarflexion—have emerged as a versatile solution for managing foot drop (Civil et al., 2023; Krishnan et al., 2024a). The elastic nature of soft AFOs allow for dorsiflexion assistance while allowing greater ankle range of motion than rigid or flexible AFOs, which could allow for more symmetrical propulsion (Totah et al., 2019). Soft AFOs are also modular and can therefore be configured with different stiffnesses to assist a broader range of impairment levels than other commercially available AFOs. Furthermore, because soft AFOs consist of softer materials like neoprene, they offer a comfortable alternative to traditional AFOs that often suffer from poor compliance due to wearer pain and discomfort (Bashir et al., 2022; Swinnen & Kerckhofs, 2015). Soft AFOs can also be worn on top of a user's shoe or barefoot, further increasing comfort. Indeed, previous studies comparing soft and traditional AFOs in clinical populations have found that soft AFOs have higher usability scores without compromising function (Civil et al., 2023). However, despite the current commercial availability of soft AFOs (e.g., FootFlexor AFO or Tenbon Adjustable AFO) and prior research examining traditional AFOs (Arazpour et al., 2013; Caliskan Uckun et al., 2014; Daryabor et al., 2018; Gasq et al., 2023), no studies have comprehensively evaluated the biomechanical effects of soft AFOs on gait and compared them with more traditional AFOs.
Here, we examined the biomechanical and neurophysiological impacts of rigid, flexible, and soft AFOs on gait. Participants walked either without assistance from an AFO (No AFO) or while wearing one of the AFOs. During walking, we measured muscle activity, kinematics, and ground reaction forces. We hypothesized that the soft AFOs would provide greater ankle angular velocity and plantarflexion while maintaining a similar level of ankle dorsiflexion support in comparison with traditional AFOs. We also hypothesized that soft AFOs would result in greater propulsive ground reaction forces and higher plantarflexor muscle activation when compared with traditional AFOs.
Materials and Methods
Participants
23 neurologically intact adults participated in this study (8 male, 15 female; age: 19.5
Experimental Apparatus and Protocol
The experiment consisted of a Static trial, a No AFO condition, and four ankle foot orthoses (AFO) conditions. Participants either stood or walked on an instrumented treadmill while a motion capture system tracked the participant's sagittal plane kinematics in real-time (Figure 1(A)). During the Static trial, participants stood in their anatomical position for 10 s to capture their neutral joint angles and body weight. During the No AFO condition, participants walked for 60 s at 1 m/s with a pair of laboratory shoes (Influence Jimmy Canvas Slip Sneakers Low Rise) and no assistance from any AFOs. During each AFO condition, participants walked on the treadmill for 60 s at 1 m/s while wearing one of the commercially available AFOs included in our study. In all AFO conditions, the AFO was worn on their dominant leg.

(A) Motion capture set-up used to measure sagittal plane kinematics. Retroreflective markers on the greater trochanter, lateral epicondyle, lateral malleolus, and foot were used to compute real-time angles of the hip, knee, and ankle. (B) Commercially available ankle-foot orthoses (AFOs) used in the current study, listed from left to right: A rigid anterior AFO (ToeOFF, Allard USA Inc., Rockaway, NJ, USA), a flexible posterior leaf spring AFO (Brace Direct Semi-Rigid AFO, Soddy-Daisy, TN, USA), a soft AFO with dorsiflexion assistance (NewGait®, Elite Athlete Products, Inc., San Diego, CA, USA), and a soft AFO with dorsiflexion + plantarflexion assistance (NewGait®, Elite Athlete Products, Inc., San Diego, CA, USA). Here, the rigid anterior AFO represents rigid AFOs designed to immobilize the ankle joint during gait and the posterior leaf spring AFO represents flexible AFOs designed to allow some mobility for comfort.
During the AFO conditions, four commercially available AFOs were evaluated in a random order: a rigid AFO, a flexible AFO, and two soft AFOs (Figure 1(B)). The rigid AFO was a rigid anterior AFO (ToeOFF, Allard USA Inc., Rockaway, NJ, USA) that immobilized the ankle with a carbon fiber structure. The flexible AFO was a posterior leaf spring (PLS) AFO (Brace Direct Leaf Spring Semi-Rigid AFO, Soddy-Daisy, TN, USA) that supported the ankle with a flexible plastic structure. The choices for rigid and flexible AFOs were based on the commonness of their prescription in clinical populations including stroke, cerebral palsy, spinal cord injury, and peripheral nerve injury (Choo & Chang, 2021a). These choices were also confirmed by consultation with a certified orthotist and prosthetist (CPO) who works with these populations. Both soft AFOs were the ankle portions of the post-stroke NewGait® (Elite Athlete Products, Inc., San Diego, CA, USA), a commercially available soft exosuit for gait rehabilitation (Krishnan et al., 2024a; Tooman, 2022; Tooman et al., 2023; Ustinova & Langenderfer, 2024). The ankle portion of the NewGait® consists of a neoprene strap worn on the user's shank just below the knee, a shoe anchor with a loop placed around the heel, and a cord fastened around the shoe near the metatarsophalangeal joints. The shank strap included attachable D-rings on the anterior and posterior surfaces, and elastic bands with carabiner ends could connect from the anterior D-ring of the shank strap to the shoe cord for dorsiflexion assistance and/or from the posterior D-ring to the shoe anchor loop for plantarflexion assistance. The shank strap was positioned above the heads of the gastrocnemius muscles to prevent the band from sliding down the shank during use. The tension in the shank strap could be adjusted to further ensure that it did not move. The NewGait® AFO included elastic bands of differing colors to denote stiffness levels (yellow, red, and green, green being the stiffest) that span the ankle to modify the level of assistance to dorsiflexion and/or plantarflexion. We examined the NewGait® AFO in two configurations: dorsiflexion assistance, and dorsiflexion + plantarflexion assistance. Dorsiflexion assistance was provided with the medium stiffness elastic band (red), unless the participant felt that the band was uncomfortably stiff, in which case we used the yellow elastic band instead. For the dorsiflexion + plantarflexion configuration, we added a medium stiffness band to assist plantarflexion and changed to a less stiff elastic band (yellow) if the participant experienced any discomfort. All AFOs were donned following manufacturer recommendations.
Data Measurement and Analysis
The motion tracking system (30 Hz, C920 Pro HD, Logitech, Newark, CA, USA) (Krishnan et al., 2015; Saner et al., 2017) monitored the positions of 19 mm retroreflective markers placed on the greater trochanter, lateral epicondyle of the femur, lateral malleolus, and the head of the fifth metatarsal. A custom LabView program (LabView 2011, National Instruments, Austin, TX, USA) with Vision Assistant (NI Vision Development Module, 2013, National Instruments, Austin, TX, USA) was used to track the participants’ real-time kinematics (sagittal plane hip, knee, and ankle joint angles) based on the positions of these markers (Figure 1(A)).
During the experiment, we measured ground reaction forces using the instrumented treadmill (Bertec Corporation, Columbus, OH, USA) and muscle activity using surface electromyography (EMG, Trigno Avanti, Delsys, Natick, MA, USA) of eight lower extremity muscles on the participant's dominant leg: vastus medialis (VM), rectus femoris (RF), medial hamstring (MH), lateral hamstring (LH), tibialis anterior (TA), medial gastrocnemius (MG), lateral soleus (Sol), and gluteus medius (GlutMed). EMG electrodes were securely attached to the participant's leg using self-adhesive tapes, 3 M Transpore medical tapes, and cotton elastic bandages. The skin above the muscle bellies was prepped prior to EMG placement using alcohol prep pads, and placement locations were standardized following SENIAM guidelines (Washabaugh et al., 2023). Prior to the Static trial, participants performed maximum voluntary contractions (MVCs) of knee extensors, knee flexors, ankle dorsiflexors, ankle plantar flexors, hip flexors, and hip abductors against manual resistance, with loud verbal encouragement to ensure maximum effort (Augenstein et al., 2024; Washabaugh et al., 2023; Yang et al., 2024). Each contraction was performed in a seated position except for the hip abduction contraction, which was performed while standing. EMG and ground reaction force signals were filtered prior to sampling using an analog low-pass Butterworth filter (500 Hz, NI SCXI-1143, National Instruments) to prevent high frequency signal aliasing and then digitally sampled at 1000 Hz using a 16-bit National Instruments Data Acquisition device (NI USB-6255).
Sampled signals were processed using a custom MATLAB program (R2019b, MathWorks, Natick, MA, USA). EMG signals were then digitally filtered using a zero-lag 20–500 Hz bandpass Butterworth filter (4th order), rectified, smoothed using a zero-lag, 6 Hz lowpass Butterworth filter (4th order) to create a linear envelope, and then normalized to the MVC. Ground reaction force and kinematic data were digitally filtered with a zero-lag, 6 Hz lowpass Butterworth filter (4th order). Following this, ankle joint velocity (reported in °/s) was computed by taking the rate of change (i.e., time-derivative) of the measured ankle joint angle. Ankle joint velocity was included because recent research has shown that it is a more reliable indicator of post-stroke dorsiflexion function during gait than the ankle joint angle (Srivastava et al., 2024). Kinematic, EMG, and ground reaction force data were then segmented into strides based on gait events (heel strike) determined using ankle marker and vertical ground reaction force data. Strides were then interpolated to 101 evenly spaced data points and ensemble averaged to create a single stride for each condition. Joint angles measured in the static posture were deducted from the joint angles measured during walking. Gait phases were defined based on previously established guidelines and known changes in gait phases based on gait speed (Dong et al., 2018): loading response (0–10%), mid-stance (10–35%), terminal stance (35–60%), pre-swing (60–70%), initial swing (70–80%), mid-swing (80–90%), and terminal swing (90–100%).
Statistics
We used one-dimensional statistical parametric mapping (spm1d) to examine changes across experimental conditions in the kinematic, muscle activation, and ground reaction force variables. spm1d is a statistical analysis method that is conceptually identical to performing analyses of variance (ANOVA) at each point in the gait cycle, except that the critical test threshold controls the family-wise error rate. Here, spm1d allowed us to perform a more informative analysis than a traditional ANOVA because traditional ANOVAs can only operate on single data points. Therefore, a traditional ANOVA would require a priori assumptions regarding which point(s) are expected to vary between conditions for each variable, or retroactive decisions following visual inspection of the data, which can introduce bias. By considering all points in the gait cycle, spm1d is able to avoid both of these potential issues (Pataky et al., 2013). For each biomechanical variable, we used spm1d to perform a repeated-measures, one-way ANOVA with walking condition (five levels: No AFO and each AFO condition) as the within-subjects factor.
A significant main effect was followed by post-hoc paired t-tests with Benjamini-Hochberg False Discovery Rate correction. Post-hoc tests were only performed on positive peaks of the kinematic, EMG, or ground reaction force data or negative peaks of the kinematic or ground reaction force data that occurred within the spm1d significant region. Therefore, each participant's positive or negative peak value in the significant region was used in the post-hoc analysis, and if no peak was present in the spm1d region then no post-hoc comparison was performed on that region. This analysis was used to perform meaningful post-hoc comparisons on spm1d regions that included both positive and negative peaks in data where positive and negative peaks have different physical meanings (e.g., braking or propulsion both appear in A/P GRF). If two peaks of the same sign occurred within the same significant region, only the first peak was considered.
Results
To improve readability, the results of this study are presented in both a table summarizing the key findings in each outcome variable (Table 1) as well as the detailed results of our statistical analysis in Kinematics, Electromyography, and Ground Reaction Force subsections.
Summary of Key Findings.
Dir: Direction of Ground Reaction Force. Soft AFO (D): Soft AFO with Dorsiflexion Assistance, Soft AFO (D + P): Soft AFO with Dorsiflexion and Plantarflexion assistance. TA: Tibialis Anterior, MG: Medial Gastrocnemius, Sol: Lateral Soleus, VM: Vastus Medialis, RF: Rectus Femoris, MH: Medial Hamstring, LH: Lateral Hamstring, GlutMed: Gluteus Medius.
Kinematics
Ankle Joint Angle and Velocity
We detected significant differences between conditions in the ankle joint angle (Figure 2(A)). Specifically, there were significant main effects of condition during loading response (0.0–13.7% of the gait cycle, p = 0.013), mid-stance to terminal stance (25.4–56.0%, p < 0.001), and pre-swing through terminal swing (59.9–100.0%, p < 0.001). Within these regions, there were four positive/negative peaks: maximum dorsiflexion during terminal stance (

(A) Ensemble average ankle joint angle (left, positive denotes dorsiflexion) and velocity (right, positive denotes dorsiflexion velocity) during the gait cycle. Shaded regions around each trace represent the standard error of the mean at each point in the gait cycle. The grey shaded regions along the horizontal axis denote significant main effects of condition detected by statistical parametric mapping (spm1d). (B) Absolute angle and velocity measurements at peaks/valleys in the regions of gait cycle where spm1d found differences between conditions. Here, “Stance Dorsiflexion” refers to the ankle angle peak at ≈50% of the gait cycle, “Swing Dorsiflexion” refers to the ankle angle peak at ≈ 80% of the gait cycle, “Stance Plantarflexion” refers to the ankle angle negative peak at ≈10% of the gait cycle, “Swing Plantarflexion” refers to the ankle angle negative peak at ≈70% of the gait cycle. “Dorsiflexion Velocity” and “Plantarflexion Velocity” refer to the peaks and valleys at ≈75% and ≈ 65%, respectively. Brackets denote significant differences between conditions. NA: No AFO, R: Rigid AFO, F: Flexible AFO, Soft (D): Soft AFO with Dorsiflexion Assistance, Soft (D+P): Soft AFO with Dorsiflexion+Plantarflexion Assistance. In each plot, each point denotes an individual participant, the lines connecting data points indicate the same participant across multiple conditions, and the error bars denote standard error of the mean. Here, we observed that while all AFOs reduced ankle plantarflexion, the Soft AFOs reduced ankle plantarflexion to a lesser extent than the rigid and flexible AFOs.
For ankle joint velocity, there were significant main effects of experimental condition during loading response (3.7–7.6%, p = 0.011), mid-stance (17.3–32.5% and 33.9–43.0%, both p < 0.001), and terminal stance through mid-swing (51.8–87.1%, p < 0.001, Figure 2(A)). Within these regions, peak plantarflexion velocity occurred during pre-swing (
Hip and Knee Joint Angles
We detected significant differences between conditions in the hip joint angle (Figure 3(A)). Specifically, there were significant main effects of condition during terminal stance (37.7–41.3% of the gait cycle, p = 0.046) and mid-swing (81.0–89.5%, p = 0.033). Within these regions, there was a visible peak in the second region (

(A) Ensemble average hip (left, positive denotes flexion) and knee (right, positive denotes flexion) joint angle during the gait cycle. Shaded regions around each trace represent the standard error of the mean at each point in the gait cycle. The grey shaded regions along the horizontal axis denote significant main effects of condition detected by statistical parametric mapping (spm1d). (B) Absolute angle measurements at peaks/valleys in the regions of gait cycle where spm1d found differences between conditions. Here, “Hip Flexion” refers to the hip angle peak at ≈86% of the gait cycle, “Knee Flexion” refers to the hip angle negative peak at ≈ 42% of the gait cycle. Brackets denote significant differences between conditions. NA: No AFO, R: Rigid AFO, F: Flexible AFO, Soft (D): Soft AFO with Dorsiflexion Assistance, Soft (D+P): Soft AFO with Dorsiflexion+Plantarflexion Assistance. In each plot, each point denotes an individual participant, the lines connecting data points indicate the same participant across multiple conditions, and the error bars denote standard error of the mean. Here, we observed that the rigid AFO reduced knee flexion during terminal stance compared to all conditions, and both soft AFOs reduced mid-swing phase hip flexion compared to No AFO.
We detected significant differences between conditions in the knee joint angle (Figure 3(A)). Specifically, there were significant main effects of condition during terminal stance (33.8–49.9% of the gait cycle, p = 0.005), pre-swing (62.9–68.3% of the gait cycle, p = 0.039) and terminal swing (79.3–92.6%, p = 0.011). Within these regions, there was a visible negative peak in the first region (
Electromyography
Ankle Muscles
We detected significant differences in activation of the tibialis anterior (TA) during the loading response (0.0–8.0% of the gait cycle, p = 0.005), mid-to-terminal stance (16.5–23.7% and 31.4–39.6%, p = 0.008 and 0.001, respectively), and swing phase (65.8–100.0%, p < 0.001, Figure 4(A)), but a peak was only present during the initial swing phase (

(A) Ensemble average tibialis anterior (TA, left), medial head of the gastrocnemius (MG, middle), and lateral soleus (Sol, right) during the gait cycle. Shaded regions around each trace represent the standard error of the mean at each point in the gait cycle. The grey shaded regions along the horizontal axis denote significant main effects of condition detected by statistical parametric mapping (spm1d). (B) Activation measurements at peaks in the regions of gait cycle where spm1d found differences between conditions. Here, “Swing Phase TA” refers to the TA activation peak at ≈70% of the gait cycle, and “Sol” refers to the Sol activation angle peak at ≈ 50% of the gait cycle. Brackets denote significant differences between conditions. NA: No AFO, R: Rigid AFO, F: Flexible AFO, Soft (D): Soft AFO with Dorsiflexion Assistance, Soft (D+P): Soft AFO with Dorsiflexion+Plantarflexion Assistance. In each plot, each point denotes an individual participant, the lines connecting data points indicate the same participant across multiple conditions, and the error bars denote standard error of the mean. Here, we observed that the soft AFOs reduced swing phase TA activation and the rigid AFO reduced Sol activation.
We detected significant differences in activation of the medial gastrocnemius (MG) during terminal stance (29.7–40.1%, p < 0.001) and initial swing (71.5–74.9%, p = 0.024), although no clear peaks were present in these regions (Figure 4(A)). We detected significant differences in activation of the Soleus (Sol) during the loading response (2.1–8.0% of the gait cycle, p = 0.039) and terminal stance (41.4–53.7%, p = 0.001), although a clear peak was only present during the terminal stance region (Figure 4(B)). Post-hoc analysis revealed that the rigid AFO significantly reduced Sol activation as compared to all conditions (all p < 0.05). Additionally, the soft AFO with dorsiflexion + plantarflexion assistance significantly reduced Sol activation relative to No AFO (p = 0.049).
Proximal Muscles
Electromyography from the more proximal lower extremity muscles (VM, RF, MH, LH, and GlutMed) are shown in Figure 5. Due to technical difficulties, RF from one participant was not included in the analysis. We detected significant differences in activation of the medial hamstring (MH) during the mid-to-terminal stance (26.5–59.6%, p < 0.001), and a visible peak was present in this region (

(A) Ensemble average vastus medialis (VM, top left), rectus femoris (RF, top middle), medial hamstring (MH, top right), lateral hamstring (LH, bottom left), and gluteus medius (GlutMed, bottom left) during the gait cycle. Shaded regions around each trace represent the standard error of the mean at each point in the gait cycle. The grey shaded regions along the horizontal axis denote significant main effects of condition detected by statistical parametric mapping (spm1d). (B) Activation measurements at peaks in the regions of gait cycle where spm1d found differences between conditions. Here, “Stance MH” refers to the MH activation peak at ≈30% of the gait cycle. Brackets denote significant differences between conditions. NA: No AFO, R: Rigid AFO, F: Flexible AFO, Soft (D): Soft AFO with Dorsiflexion Assistance, Soft (D+P): Soft AFO with Dorsiflexion+Plantarflexion Assistance. In each plot, each point denotes an individual participant, the lines connecting data points indicate the same participant across multiple conditions, and the error bars denote standard error of the mean. Here, we observed that the rigid AFO increased MH activation during mid-stance.
Ground Reaction Forces
We detected significant differences in mediolateral (M/L) ground reaction forces (GRFs) during the loading response (4.1–10.5%, p = 0.009), mid-stance (19.4–41.1%, p < 0.001) and pre-swing phases (59.7–66.9%, p = 0.006, Figure 6(A)). We observed a force peak during the mid-stance region (
We detected significant differences in anteroposterior (A/P) GRFs during the loading response through mid-stance (97.0–0.4% and 4.3–34.5%, p = 0.020 and < 0.001, respectively), terminal stance (50.3–58.1%, p < 0.001), and pre-swing phase (63.2–69.4%, p = 0.001) (Figure 6(A)). We observed a braking force peak during the mid-stance region and a propulsion force peak during the terminal stance region. Post-hoc analysis of the braking force peak revealed that the rigid AFO increased braking relative to No AFO, flexible AFO, and soft AFO with dorsiflexion assistance (all p < 0.040, Figure 6(B)). Similarly, the soft AFO with dorsiflexion + plantarflexion assistance increased the braking force peak relative to the soft AFO with dorsiflexion assistance (both p < 0.05). Post-hoc analysis of the propulsion force peak revealed that the rigid and flexible AFOs reduced propulsion relative to No AFO and both soft AFOs (all p < 0.05) but did not differ from each other. We did not detect differences in propulsion between No AFO and the soft AFOs. We detected significant differences in vertical GRFs during the loading response through mid-stance (3.3–15.3%, p < 0.001), terminal stance (39.2–46.5%, p < 0.001), and pre-swing phase (57.1–68.3%, p = 0.001), although none of these regions contained a peak (Figure 6).

(A) Ensemble average mediolateral (left, positive denotes medial), anteroposterior (A/P) (middle, positive denotes anterior force), and vertical (right) ground reaction forces during the gait cycle. Shaded regions around each trace represent the standard error of the mean at each point in the gait cycle. The grey shaded regions along the horizontal axis denote significant main effects of condition detected by statistical parametric mapping (spm1d). (B) Force measurements at peaks/valleys in the regions of gait cycle where spm1d found differences between conditions. Here, “Lateral Force” refers to the force peak at ≈35% of the gait cycle, “Brake Force” refers to the force peak at ≈20% of the gait cycle, and “Propulsive Force” refers to the force valley at ≈55% of the gait cycle. Brackets denote significant differences between conditions. Brackets denote significant differences between conditions. NA: No AFO, R: Rigid AFO, F: Flexible AFO, Soft (D): Soft AFO with Dorsiflexion Assistance, Soft (D+P): Soft AFO with Dorsiflexion+Plantarflexion Assistance. In each plot, each point denotes an individual participant, the lines connecting data points indicate the same participant across multiple conditions, and the error bars denote standard error of the mean. Here, we observed that the rigid and flexible AFOs reduced the peak propulsive force.
Discussion
The purpose of this study was to comprehensively evaluate changes in gait biomechanics and muscle activation due to rigid, flexible, and soft ankle foot orthoses (AFOs). Neurologically intact participants walked with four different commercially available AFOs (one rigid, one flexible, and two soft), and we compared their effects on sagittal plane kinematics, electromyography, and ground reaction forces. We found that rigid (ToeOFF), flexible (Brace Direct Semi-Rigid AFO), and soft AFOs (the NewGait®) had differential effects on ankle joint angles and velocity, propulsive ground reaction forces, and activation of ankle muscles. Specifically, we found that all AFOs reduced peak plantarflexion angle and velocity when compared with No AFO, but the soft AFOs increased ankle dorsiflexion velocity, plantarflexion velocity, and plantarflexion angle while retaining swing phase dorsiflexion compared to the rigid and flexible AFOs. We also found that the soft AFOs increased propulsion compared to both the rigid and flexible AFOs. These findings indicate that soft AFOs allow greater plantarflexor propulsion than other AFOs while preserving dorsiflexion assistance and highlight differences between conventional AFOs and the potential advantages of commercially available soft AFOs. While these findings were from neurologically intact individuals, they may have implications for both the clinicians who prescribe AFOs and the patient populations who wear them, such as stroke, cerebral palsy, or spinal cord injury.
While the AFOs in this study often produced similar effects on gait biomechanics (e.g., reduced peak plantarflexion angle and velocity during push-off/pre-swing), there were several notable differences between AFOs arising from how each AFO supported the ankle. For example, the soft AFOs generally interfered with the ankle's contribution to propulsion to a lesser extent than the rigid and flexible AFOs. This conclusion is reinforced by the rigid and flexible AFOs reducing plantarflexion ankle velocity at push-off/pre-swing and the propulsive force peak to a greater extent than the soft AFOs. This advantage of greater propulsion from the ankle with the soft AFOs is likely due to the elasticity of the soft AFOs allowing greater range of motion than the rigid and flexible AFOs. However, it is critical to point out that a primary function of an AFO is to prevent foot drop in addition to other useful functions such as providing stability, improving balance and mobility, and ensuring toe/limb clearance. Therefore, the finding that soft AFOs allow greater plantarflexion movement during push-off is only meaningful if it does not compromise dorsiflexion assistance during the swing phase. To this end, we also observed greater ankle dorsiflexion velocity during the initial swing phase and a sustained ankle dorsiflexion angle during swing phase equal to or greater than the rigid and flexible AFOs. This indicates that the greater range of motion afforded by the soft AFOs did not compromise the necessary function of an AFO.
It is important to note that our results highlight that, while restricting ankle motion is a necessary result of a passive AFO to support ankle dorsiflexion, an overly restrictive AFO can produce unintended gait alterations. For instance, the rigid AFO showed the greatest reduction in plantarflexion velocity at push-off/pre-swing as well as a greater reduction in peak propulsive force than both soft AFOs. This reduction in ankle propulsion resulted in compensatory neurophysiological and kinematic changes both at and above the ankle. While wearing the rigid AFO, participants reduced their peak lateral soleus activation and increased their knee extension and medial hamstring activation [a knee flexor/hip extensor] during terminal stance (Figures 3–5). This could be a compensatory strategy to increase reliance on hip extensors for propulsion. The combination of these results with the aforementioned propulsive advantages of soft AFOs suggest that soft AFOs could be an appropriate option for individuals who suffer from foot drop but do not require full ankle immobilization. These findings provide insight into how different AFOs affect walking biomechanics as well as the potential advantages of soft AFOs over rigid AFOs, which may be instructive to rehabilitation engineers designing new AFOs for foot drop. However, it is important to note that our findings were in neurologically intact individuals and findings may differ in similar studies with clinical populations (e.g., stroke, MS). Therefore, future work should examine the biomechanical effects of soft AFOs in these populations before designing or making clinical recommendations.
While no study has yet, to our knowledge, performed a comprehensive biomechanical comparison of rigid, flexible, and soft AFOs, we can compare our results with studies that examine how AFOs affect gait biomechanics. The results of our study are consistent with prior studies, which show that AFO assistance to ankle dorsiflexion can limit both ankle plantarflexion at push-off and overall range of motion (Kobayashi et al., 2013; Mulroy et al., 2010; Totah et al., 2019). We can further compare our results with studies that examine the effects of AFO stiffness on gait biomechanics. These studies typically examine a hinged AFO with a configurable spring constant at the hinge and vary the spring constant from low to high stiffness (Choi et al., 2017; Waterval et al., 2019). In the context of our study, the soft, flexible, and rigid AFOs could be considered as “low”, “medium”, and “high” stiffness conditions, respectively. Like our study, these previous studies also found that increasing stiffness reduced peak stance phase dorsiflexion angle and plantarflexor activation during pre-swing (Choi et al., 2017; Waterval et al., 2019). Similarly, previous studies that have examined AFOs that provide unidirectional spring assistance to dorsiflexion (like the soft AFO with dorsiflexion assistance in the current study) have found that these AFOs reduce peak plantarflexion and increase peak dorsiflexion in late stance but do not reduce propulsive ground reaction forces (Yamamoto et al., 2020), similar to our findings.
The advantages of soft AFOs shown in this study may have consequences for clinical populations (e.g., stroke, multiple sclerosis). Specifically, while previous studies have shown that AFOs can improve gait in clinical populations, many individuals feel that AFOs can overly restrict their ankle joint motion and make them feel unsteady (Nahorniak et al., 1999; Sienko Thomas et al., 2002; Totah et al., 2019; van der Wilk et al., 2018). These individuals will instead opt for devices like peroneal nerve stimulators that prevent foot drop while allowing more naturalistic ankle excursion and greater propulsion with their paretic limb (Sheffler et al., 2006; van Swigchem et al., 2010). However, these stimulators are more expensive than passive AFOs and are not always covered by third-party insurance, reducing their accessibility to a large portion of clinical populations. Soft AFOs, however, may represent a low-cost alternative to these stimulators, as our results show that they allow for greater ankle excursion and limb propulsion while still assisting in foot drop. This greater ankle excursion could have allowed the ankle plantarflexors to contribute greater propulsive power, which is critical for maintaining gait speed in neurological populations (Awad et al., 2020; Jonkers et al., 2009; Nadeau et al., 1999). Indeed, findings from a recent human-centered design study involving stroke survivors and the NewGait® system (which consists of the soft AFOs) indicate that participants felt that the system dramatically improved their stability (Krishnan et al., 2024a). Furthermore, we collected subjective feedback from the participants following walking in the current study to compare each AFO based on participant preferences. In general, most participants preferred one of the soft AFOs due to the comfort and feel of the elastic tension. Few participants preferred the rigid or flexible AFOs because they felt uncomfortable or overly restricted (see Supplemental Table 1). However, it is important to note that the subjective AFO preferences of neurologically intact adults, who do not require an AFO to walk safely, may differ from clinical populations. For example, one key weakness of the soft AFOs in clinical populations with upper extremity impairment (e.g., stroke) may be that the elastic bands make donning and doffing soft AFOs difficult. Therefore, future work with soft AFOs in clinical populations should not only examine biomechanical and neurophysiological differences between AFOs, but also collect subjective feedback comparing comfort, stability, and usability.
In summary, the results of this study comprehensively establish the effect of different AFOs (rigid, flexible, and soft) on gait biomechanics and lower extremity muscle activation. Our results indicate that, in general, all AFOs reduced ankle joint excursion and plantarflexion velocity during gait and the performance of AFO improved with reducing stiffness (see Table 1 for a full summary of key findings). Particularly, the soft AFOs allowed for greater ankle excursion and velocity, plantarflexor activation, and propulsive force without compromising foot drop assistance during the swing phase. We believe that these findings offer critical information regarding the differences between commercially available AFOs and may have implications for the adoption of AFOs in clinical populations.
Supplemental Material
sj-docx-1-rnn-10.1177_09226028251395707 - Supplemental material for Soft vs. Traditional AFOs: A Comparative Study on Gait Kinematics, Kinetics, and Muscle Activity
Supplemental material, sj-docx-1-rnn-10.1177_09226028251395707 for Soft vs. Traditional AFOs: A Comparative Study on Gait Kinematics, Kinetics, and Muscle Activity by Thomas E. Augenstein, Breanna Bordine, Nyari Bhatt, Shreeya Buddaraju, Olugbenga P. Adeeko, Edward P. Washabaugh and Chandramouli Krishnan in Restorative Neurology and Neuroscience
Footnotes
Ethical Considerations
This study received ethical approval from the University of Michigan Medical School IRB (IRBMED HUM00093673).
Consent to Participate
All participants provided written informed consent to participate in the study.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Eunice Kennedy Shiver National Institute of Child Health and Human Development / National Institutes of Health (R41-HD111289).
Declaration of Conflicting Interests
OPA is the inventor of the original NewGait device and CEO/Founder of the company that developed and sells the product. The authors declare no additional potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data Availability
The data supporting the findings of this study are available on reasonable request to the corresponding author.
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
