Journal of NeuroEngineering and Rehabilitation 22, Article number: 196 (2025)
Effective ankle plantarflexion assistance (PF assistance) requires precise timing and personalized adjustment using an ankle joint robot that assists both dorsiflexion and plantarflexion during walking. Previous studies suggest that peak paretic propulsion timing (Peak PP timing) is earlier in some individuals, indicating differences in ground reaction force (GRF) characteristics among individuals with stroke. Therefore, this study aimed to identify the effective timing of PF assistance according to anteroposterior GRF characteristics.
The study included twenty-three post-stroke individuals with hemiparesis. The participants walked on a treadmill under an unassisted condition, and at 30–50% and 40–60% gait cycle (GC) PF assisted conditions. The participants were clustered into two groups based on the waveform data of the paretic anteroposterior GRF in the unassisted condition using k-means clustering, and differences in the characteristics between the two groups, and changes in paretic propulsion in each assistance condition within the cluster were analyzed. Additionally, the relationship between changes in paretic propulsion and the timing of paretic propulsion (PP timing) and Peak PP timing, were analyzed within each cluster, relative to the unassisted condition, under each assistance condition.
Participants were clustered into Clusters A (N = 13) and B (N = 10). Cluster A showed earlier Peak PP timing and increased paretic propulsion in both PF assistance conditions, while Cluster B only improved in the 40–60% GC condition. A significant negative correlation was observed between changes in paretic propulsion and PP timing in Cluster A under the 30–50% GC PF assistance condition (ρ = − 0.665, P = 0.012).
Our results suggest that applying PF assistance, particularly during the late paretic stance phase, might be effective for individuals with hemiparesis, as it led to improvements in paretic propulsion in both clusters. Additionally, the results suggest that for participants in Cluster A, it may be important to apply PF assistance at the 30–50% GC while monitoring changes in the GRF waveform data, as this approach may be more effective for some participants in improving paretic propulsion.
Journal of NeuroEngineering and Rehabilitation volume 22, Article number: 9 (2025)
Biomechanical gait impairments, such as reduced paretic propulsion, are common post-stroke. Studies have used biofeedback to increase paretic propulsion and reduce propulsion asymmetry, but it is unclear if these changes impact overall gait asymmetry. There is an implicit assumption that reducing propulsion asymmetry will improve overall gait symmetry, as paretic propulsion has been related to numerous biomechanical impairments. However, no work has investigated the impact of reducing propulsion asymmetry on overall gait asymmetry. We aimed to understand how within-session changes in propulsion asymmetry affect overall gait asymmetry in individuals post-stroke, operationalized as the combined gait asymmetry metric (CGAM). We hypothesized that decreasing propulsion asymmetry would reduce CGAM. Methods. Participants completed twenty minutes of biofeedback training designed to increase paretic propulsion. We calculated the change in propulsion asymmetry magnitude (Δ|PA|) and the change in CGAM (ΔCGAM) during biofeedback relative to baseline. Then, we fit a robust linear mixed-effects model with ΔCGAM as the outcome and a fixed effect for Δ|PA|. Results. We found a positive association between Δ|PA| and ΔCGAM (β = 2.6, p = 0.002). The average Δ|PA| was -0.09, suggesting that, on average, we would expect a CGAM change of 0.2, which is 0.5% of the average baseline CGAM value. Conclusions. Reducing propulsive asymmetry using biofeedback is unlikely to produce substantial reductions in overall gait asymmetry, suggesting that biofeedback-based approaches to reduce propulsion asymmetry may need to be combined with other interventions to improve overall gait asymmetry. Clinical Trial Registration. NCT04411303.
Individuals post-stroke commonly present with biomechanical gait impairments such as increased spatiotemporal asymmetries, reduced paretic propulsion, and reduced swing knee flexion [1, 2]. Paretic propulsion is a popular target for clinical interventions and research studies because it is associated with walking speed in individuals post-stroke [3,4,5,6] and can be increased using a variety of interventions (gait biofeedback [7], functional electrical stimulation [8, 9], robotics [10], etc.). Additionally, repeated training targeting paretic propulsion can improve functional balance, walking speed [8], and cost of transport [11]. Research studies investigating these interventions mainly focus on quantifying the changes in paretic propulsion or propulsion asymmetry and do not consider whether changes in paretic propulsion impact symmetry in other gait impairments. The assessment of stroke survivor stakeholder values indicates that overall gait asymmetry (i.e., gait appearance) is a priority to address during rehabilitation [12], making it important to understand how an intervention impacts overall gait asymmetry.
There is an implicit assumption that reducing propulsion asymmetry will result in improved symmetry in the entire gait pattern, as paretic propulsion has been related to numerous biomechanical impairments such as paretic knee flexion [13], paretic and non-paretic trailing limb angle [14], and paretic step length [3]. However, no work has directly investigated the impact of reducing propulsion asymmetry on overall gait asymmetry. With numerous degrees of freedom in the lower limb, it is possible that reducing propulsion asymmetry may not reduce other biomechanical impairments and, therefore, may not improve overall gait asymmetry.
The primary aim of this study was to understand how within-session changes in propulsion asymmetry affect overall kinematic and spatiotemporal gait asymmetry in individuals with chronic stroke, operationalized by the combined gait asymmetry metric (CGAM) [15, 16]. This is a first step to understanding what kinematic and spatiotemporal changes might be expected if propulsion asymmetry was reduced in the clinic through biofeedback or other interventions. The CGAM provides a single comprehensive measure of overall kinematic and spatiotemporal gait asymmetry (bounded between 0 and 200), allowing for the inclusion of any biomechanical impairment [15, 16]. With CGAM, we can assess the impact of changing propulsion asymmetry on overall gait asymmetry, not just on a single biomechanical impairment. To manipulate propulsion asymmetry, we used visual biofeedback to increase paretic propulsion. Because of paretic propulsion’s relationship with numerous biomechanical impairments [3, 13, 14], we hypothesized that a decrease in propulsion asymmetry would reduce CGAM.
More at link.
Journal of NeuroEngineering and Rehabilitation volume 20, Article number: 85 (2023)
Individualized, targeted, and intense training is the hallmark of successful gait rehabilitation in people post-stroke. Specifically, increasing use of the impaired ankle to increase propulsion during the stance phase of gait has been linked to higher walking speeds and symmetry. Conventional progressive resistance training is one method used for individualized and intense rehabilitation, but often fails to target paretic ankle plantarflexion during walking. Wearable assistive robots have successfully assisted ankle-specific mechanisms to increase paretic propulsion in people post-stroke, suggesting their potential to provide targeted resistance to increase propulsion, but this application remains underexamined in this population. This work investigates the effects of targeted stance-phase plantarflexion resistance training with a soft ankle exosuit on propulsion mechanics in people post-stroke.
We conducted this study in nine individuals with chronic stroke and tested the effects of three resistive force magnitudes on peak paretic propulsion, ankle torque, and ankle power while participants walked on a treadmill at their comfortable walking speeds. For each force magnitude, participants walked for 1 min while the exosuit was inactive, 2 min with active resistance, and 1 min with the exosuit inactive, in sequence. We evaluated changes in gait biomechanics during the active resistance and post-resistance sections relative to the initial inactive section.
Walking with active resistance increased paretic propulsion by more than the minimal detectable change of 0.8 %body weight at all tested force magnitudes, with an average increase of 1.29 ± 0.37 %body weight at the highest force magnitude. This improvement corresponded to changes of 0.13 ± 0.03 N m kg− 1 in peak biological ankle torque and 0.26 ± 0.04 W kg− 1 in peak biological ankle power. Upon removal of resistance, propulsion changes persisted for 30 seconds with an improvement of 1.49 ± 0.58 %body weight after the highest resistance level and without compensatory involvement of the unresisted joints or limb.
Targeted exosuit-applied functional resistance of paretic ankle plantarflexors can elicit the latent propulsion reserve in people post-stroke. After-effects observed in propulsion highlight the potential for learning and restoration of propulsion mechanics. Thus, this exosuit-based resistive approach may offer new opportunities for individualized and progressive gait rehabilitation.
Stroke is a leading cause of motor disability, with over 100 million survivors worldwide [1]. Of these survivors, over 80% are left with locomotor dysfunction [2], resulting in slow and asymmetric gait presentations [3]. The incidence of stroke is projected to continue increasing over the next few decades [4], and thus presents an imminent challenge for independence and quality of life [5] for members of our communities. Reduced propulsive force generated by the paretic, or more affected, limb is a major contributor to these impairments, and leads to the inability of the individual to effectively propel the body forward [6]. This reduced paretic propulsion is partially due to weakness in the paretic ankle plantarflexor muscles [7], which leads to reduced ankle torque production, a key driver of propulsion [8]. Consequently, there is growing interest in rehabilitation programs that aim to increase paretic propulsion by targeting ankle function during the stance phase, when forward propulsion is generated, towards achieving the functional outcome of increased gait speed [9].
Among methods that improve speed and propulsion, those that elicit the latent propulsion reserve through high-intensity training have been shown to be particularly promising [10]. The presence of latent propulsion reserve is typically demonstrated as an increase in an individual’s propulsion while increasing the difficulty of the locomotor task, such as by increasing surface inclination [11] or resisting the entire body during walking [12]. However, this extra propulsion can be generated by ankle-level mechanisms (i.e., ankle kinetics) or limb-level mechanisms (i.e., proximal kinematics) [8]. Simulations suggest that traditional methods of engaging the latent propulsion reserve, such as through passive resistive elements acting on the patient’s limbs (e.g., elastic bands attached at the pelvis opposing forward motion, or weights added to the foot) [12,13,14], typically result in larger involvement of the proximal joints and affect the entire gait cycle rather than targeting the ankle in stance [15]. Over the past two decades, several wearable robotic systems for assisting the ankle during walking have demonstrated the ability to increase paretic propulsion [16] through ankle-specific mechanisms [17]. Based on the principles of high-intensity and task-specific training [18], a wearable robotic system that resists the ankle during walking may be an important approach, particularly for patients with higher propulsive capacities. Currently, however, targeted resistance training of the paretic ankle plantarflexors during stance for people post-stroke has yet to be investigated.
Most robotic systems developed for resistance training emulate conventional methods [19,20,21,22], resulting in a lack of specificity to paretic propulsion. Wearable devices offer the capability to provide controlled torques to target a specific joint and phase in the gait cycle. Consequently, some groups have developed systems for targeted swing-phase resistance in post-stroke [23] and healthy populations [24, 25], and have shown adaptations in joint kinematics that indicate increased ankle use. More recent work has shown that targeted stance-phase resistance can increase plantarflexor muscle activity in people with cerebral palsy [26] and healthy individuals [27, 28]. However, people post-stroke present with gait biomechanics and adaptation responses to perturbations that are different from both of these populations [29,30,31,32]. Thus, there is a need to explore the use of a wearable resistive robotic system for increasing paretic propulsion in stance for people post-stroke.
One challenge for developing a resistive paradigm with a wearable device is identifying the appropriate parameters of resistance. Prior literature has shown the sensitivity of users to the magnitude of active ankle resistance in able-bodied individuals [28] and passive resistance in post-stroke individuals [33]. For example, excessive resistance can lead to compensatory gait patterns that increase use of the unresisted proximal joints or limb, as evidenced by changes in limb loading or joint kinematics. Perspectives from the challenge point theory [34, 35] further support the importance of individualizing the challenge level during training to maximize retention of the learned task. Thus, there is a need for structured investigation of the effects of resistance parameters on post-stroke gait response to stance-phase ankle resistance.
An effective resistance training paradigm is one that induces learning of increased ankle use towards generating propulsion. Evidence of learning in the motor learning field is often obtained from after-effects in the few steps immediately following a perturbation [25, 27, 36], representing the persistence of an individual’s adapted state [37]. However, measuring after-effects following exoskeleton-based training has traditionally been challenging due to the added distal inertia of rigid devices, which requires a user to first doff the device, and thus may prevent capturing newly learned gait patterns. By design, the cable-driven soft exosuit only consists of textile components at the distal end of the leg, and hence can be rapidly commanded to apply no forces by releasing tension in the cables (< 50ms) [38]. In this “slack” mode, the device is transparent to the user, resulting in similar kinematics and energetics to when walking without any device [39, 40]. This transparency allows for the measurement of gait immediately after resistance without stopping walking. This approach has been used to measure changes in ankle kinematics in healthy individuals after ankle-targeted resistance [28], but has yet to be applied to people post-stroke.
In this work, we leverage a soft, cable-driven, unilateral ankle exosuit [41] to investigate the biomechanical effects of targeted stance-phase ankle resistance across varying force magnitudes in chronic survivors of stroke. We hypothesized that with this targeted approach, we would engage individuals’ latent propulsion reserve through ankle-specific mechanisms, such as ankle kinetics and plantarflexor muscle activity. We expected to observe after-effects of increased propulsion compared to baseline for strides immediately following removal of the resistive force, due to the trained increase in ankle use. We also posited that with increased force magnitude, we would observe greater gains in propulsion metrics following resistance, but at the cost of increased use of the unresisted proximal joints and non-paretic limb, based on our prior work in healthy individuals [28]. To control for the effects of speed on joint kinetics and kinematics, we conducted this investigation on a treadmill with fixed walking speeds for each individual. We performed one additional proof-of-concept exploratory experiment to assess the value of an exosuit for resistive training in which individuals walked on a treadmill without any active resistance to quantify improvements in propulsion solely from treadmill training.
More at link.
Journal of NeuroEngineering and Rehabilitation volume 20, Article number: 37 (2023)
Paretic propulsion [measured as anteriorly-directed ground reaction forces (AGRF)] and trailing limb angle (TLA) show robust inter-relationships, and represent two key modifiable post-stroke gait variables that have biomechanical and clinical relevance. Our recent work demonstrated that real-time biofeedback is a feasible paradigm for modulating AGRF and TLA in able-bodied participants. However, the effects of TLA biofeedback on gait biomechanics of post-stroke individuals are poorly understood. Thus, our objective was to investigate the effects of unilateral, real-time, audiovisual TLA versus AGRF biofeedback on gait biomechanics in post-stroke individuals.
Nine post-stroke individuals (6 males, age 63 ± 9.8 years, 44.9 months post-stroke) participated in a single session of gait analysis comprised of three types of walking trials: no biofeedback, AGRF biofeedback, and TLA biofeedback. Biofeedback unilaterally targeted deficits on the paretic limb. Dependent variables included peak AGRF, TLA, and ankle plantarflexor moment. One-way repeated measures ANOVA with Bonferroni-corrected post-hoc comparisons were conducted to detect the effect of biofeedback on gait biomechanics variables.
Compared to no-biofeedback, both AGRF and TLA biofeedback induced unilateral increases in paretic AGRF. TLA biofeedback induced significantly larger increases in paretic TLA than AGRF biofeedback. AGRF biofeedback increased ankle moment, and both feedback conditions increased non-paretic step length. Both types of biofeedback specifically targeted the paretic limb without inducing changes in the non-paretic limb.
By showing comparable increases in paretic limb gait biomechanics in response to both TLA and AGRF biofeedback, our novel findings provide the rationale and feasibility of paretic TLA as a gait biofeedback target for post-stroke individuals. Additionally, our results provide preliminary insights into divergent biomechanical mechanisms underlying improvements in post-stroke gait induced by these two biofeedback targets. We lay the groundwork for future investigations incorporating greater dosages and longer-term therapeutic effects of TLA biofeedback as a stroke gait rehabilitation strategy.
Trial registration NCT03466372
Hemiparesis following stroke causes unilateral deficits in gait kinematics and kinetics, contributing to slowed gait speed, gait asymmetries, and increased fall risk [1,2,3]. While increasing gait speed is a major goal of stroke rehabilitation [4, 5], improvements in speed can be achieved either through restoration of paretic limb function or compensatory strategies [6]. Measurement of kinematic and kinetic gait biomechanics variables can parse out restoration versus compensation as sources of gait recovery or training-induced improvements [7]. Reduced paretic propulsion, measured as the anterior component of the ground reaction force (AGRF) generated during late stance, is an important biomechanical deficit closely associated with gait speed and walking function post-stroke [8,9,10,11]. Importantly, individuals post-stroke demonstrate a paretic propulsive reserve [12] that can be exploited using gait training interventions [13, 14], with improvements in propulsion correlating to improvements in gait speed [8]. Thus, propulsion has emerged as a key modifiable post-stroke gait variable that is biomechanically and clinically relevant.
Previous studies have demonstrated two major biomechanical gait variables that contribute to overall propulsion, ankle plantarflexor moment, and trailing limb angle [15]. Ankle plantarflexors generate most of the force required to facilitate a smooth stance-to-swing transition [16]. Trailing limb angle (TLA), a measure of the overall limb angle or position with respect to the center of mass, places the leg in a better orientation to direct ground reaction forces more anteriorly [17]. Individuals post-stroke demonstrate deficits in both paretic plantarflexor moment and TLA [15], yet increases in paretic propulsion appear to originate mainly from improvements in paretic TLA [12, 18]. Post-stroke AGRF and TLA show robust inter-relationships, indicating that TLA can be used as a surrogate for paretic AGRF measurements [19]. Taken together, these studies suggest that targeting post-stroke TLA deficits may be a feasible and effective way of improving paretic propulsion.
Real-time biofeedback has emerged as a promising post-stroke gait training strategy that can target specific gait deficits on the paretic limb [20,21,22,23]. Previously, unilateral biofeedback targeting paretic propulsion was shown to induce significant increases in paretic limb propulsion without concomitant compensatory changes in the non-paretic limb [20]. However, translation of propulsion biofeedback from laboratory to clinic remains difficult because laboratory-based instrumented walkways or treadmills needed to measure AGRF may not be clinically accessible. The use of portable and wearable AGRF sensors for gait assessment is under study and not yet clinically available [24]. Moreover, estimation of propulsion through observational gait analysis is challenging even for movement experts with considerable clinical experience [25]. In contrast, measurements of TLA do not require the use of force platforms, and can be more easily subjectively estimated by clinicians based on the relationship of the forefoot to the pelvis or greater trochanter during observational gait analysis [19]. Recently, we demonstrated that able-bodied individuals are able to modulate TLA and AGRF unilaterally in response to real-time unilateral TLA biofeedback [26]. Thus, TLA biofeedback holds promise as a clinically applicable intervention that could preferentially increase paretic AGRF and reduce post-stroke propulsion deficits. While AGRF and TLA have been studied together as outcome variables in previous research, to our knowledge, biofeedback for these 2 biomechanical targets has not been directly compared in people post-stroke. Thus, an initial assessment of the feasibility and immediate biomechanical effects of TLA biofeedback on post-stroke individuals is needed.
Here, we studied the effects of TLA biofeedback on post-stroke gait biomechanics. Moreover, to assess its use as a suitable and clinically applicable alternative to AGRF biofeedback, we compared the immediate biomechanical effects of TLA biofeedback to AGRF biofeedback. We hypothesized that a biofeedback paradigm targeting paretic TLA would elicit favorable improvements in paretic propulsion and other post-stroke gait biomechanics impairments that are comparable in magnitude to AGRF biofeedback.
More at link.
