Independent and synergistic effects of visual biofeedback and phase-specific belt deceleration on affected-leg propulsion in post-stroke split-belt treadmill training

Independent and synergistic effects of visual biofeedback and phase-specific belt deceleration on affected-leg propulsion in post-stroke split-belt treadmill training | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Independent and synergistic effects of visual biofeedback and phase-specific belt deceleration on affected-leg propulsion in post-stroke split-belt treadmill training Seoung Hoon Park, Chihyeong Lee, Hyunje Park, Jooeun Ahn, Beom-Chan Lee This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8159960/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 9 You are reading this latest preprint version Abstract Background. Post-stroke walking deficits are closely associated with impaired forward propulsion from the affected leg. We developed a real-time training system using an instrumented split-belt treadmill that provides phase-specific adaptive belt speed modulation (propulsion-facilitating mode, PF) during late stance and visual biofeedback (VB) of the affected leg’s propulsive (anterior ground reaction) force. This study investigated the effects of VB, PF, and VB + PF on gait propulsion, kinematics, and muscle activity in individuals post-stroke. Methods. Thirteen adults with chronic hemiparetic stroke completed three randomized, counterbalanced treadmill trials: VB alone, PF alone, and VB + PF. Each trial consisted of the baseline (30 steps, without intervention), the training (100 steps, assigned modality), and the post-training (30 steps, without intervention) periods. All outcome measures were obtained from the affected leg, including peak propulsive force, stride length, peak knee and ankle extension, and electromyographic activity of the major ankle and knee extensors (i.e., medial gastrocnemius, soleus, vastus medialis, and rectus femoris). Results. Significant main effects of modality, period, and their interaction were observed for all outcome measures (P < 0.0001). During the training period, VB + PF resulted in the greatest improvements across all outcome measures (i.e., peak propulsive force, stride length, peak knee and ankle extension, and electromyographic activity in the ankle and knee extensor muscles) compared with either VB or PF alone. PF alone demonstrated moderate improvements, whereas VB alone showed smaller yet statistically significant improvements. Regardless of modality, the improvements achieved during the training period were retained during the post-training period. Conclusion. In individuals post-stroke, combining real-time visual biofeedback with phase-specific adaptive belt speed modulation resulted in the greatest and most substantial improvements in affected-leg propulsion, lower-limb kinematics, and electromyographic activity in the ankle and knee extensor muscles. These findings demonstrate the effectiveness of multimodal, adaptive treadmill training in engaging the affected-leg propulsion reserve and inform the development of scalable rehabilitation protocols that integrate perceptual feedback with task-specific mechanical facilitation for stroke gait rehabilitation. post-stroke gait gait propulsion adaptive belt speed modulation visual biofeedback split-belt treadmill motor learning Figures Figure 1 Figure 2 Figure 3 Figure 4 1. BACKGROUND Stroke remains one of the most prevalent causes of gait dysfunction, frequently leading to substantial impairments in ambulatory capacity and overall functional mobility [ 1 ]. These impairments are commonly characterized by reduced walking speed, compromised postural control, gait asymmetry, and increased metabolic cost, all of which are closely associated with increased risks of falls, musculoskeletal strain, early-onset fatigue, and a significant decline in quality of life [ 2 – 4 ]. Among the various biomechanical deficits observed in individuals post-stroke, impairments in push-off mechanics during the late stance phase (i.e., from mid push-off to toe-off) are particularly consequential. These impairments reduce the generation of anterior ground reaction force (AGRF), a key determinant of forward propulsion that facilitates body progression and supports symmetrical step advancement [ 2 ]. Given these impairments, restoring AGRF, particularly from the affected leg, has emerged as a central therapeutic target in gait rehabilitation. Indeed, achieving a stable and efficient gait in individuals post-stroke requires sufficient and coordinated generation of AGRF from both legs, with particular emphasis on the affected leg. Inadequate generation of AGRF in the affected leg leads to reduced propulsive output, asymmetric step transitions, and increased reliance on compensatory mechanisms, ultimately compromising gait stability and performance. Previous studies have reported that individuals post-stroke generate only about 16 to 49% of the total propulsive force using the affected leg [ 5 ]. More recently, data from a cohort of 100 stroke survivors demonstrated a significant positive association between increased propulsion from the affected leg and improved walking speed, underscoring the clinical relevance of this parameter [ 6 ]. Notably, emerging evidence suggests that individuals post-stroke retain a latent capacity to generate greater propulsive force from the affected leg [ 7 ]. This phenomenon, referred to as the affected leg propulsion reserve, reflects an underutilized capacity that may be recruited to facilitate functional recovery. Taken together, these findings emphasize that restoring forward propulsion, particularly through targeted enhancement of AGRF in the affected leg, represents a critical therapeutic objective in post-stroke gait rehabilitation. To address propulsion-related deficits, various conventional rehabilitation approaches have been implemented. These include manual therapeutic interventions such as therapeutic exercise, functional task-specific training, and the motor relearning program, which aim to restore normal movement patterns through therapist-guided facilitation and practice. Traditional interventions primarily focus on improving muscular strength, flexibility, balance, and motor control through repetitive, structured, and cognitively engaging exercises [ 8 , 9 ]. However, such conventional manual therapeutic approaches (e.g., neuro-developmental handling and facilitation-based gait training) provide limited direct enhancement of the anterior ground reaction force (AGRF), a key biomechanical determinant of forward propulsion [ 2 , 10 ]. Rather than explicitly strengthening the propulsive capacity of the affected leg, these approaches emphasize gait symmetry, joint mobility, and postural alignment. Consequently, improvements in walking speed or step length are often attributable to compensatory mechanisms rather than restoration of forward propulsion mechanics [ 2 , 10 ]. The limited efficacy of manual interventions in enhancing AGRF has motivated the exploration of more targeted, task-specific approaches. Other studies have specifically targeted the enhancement of forward propulsion by encouraging increased generation of anterior ground reaction force (AGRF) through treadmill-based gait training. One commonly studied method utilizes split-belt treadmill protocols that apply asymmetrical belt speeds (for example, a 2:1 speed ratio), with the goal of reducing gait asymmetries in individuals post-stroke [ 11 – 13 ]. A recent meta-analysis reported that this type of training can improve interlimb symmetry, and in some instances, the benefits are maintained beyond the intervention period [ 14 ]. Additionally, other studies have examined training conditions that increase the biomechanical demands of propulsion, including walking against backward-directed resistance [ 15 ] or walking on an inclined surface [ 16 ]. These interventions, while biomechanically demanding, often lack adaptive responsiveness to individual gait patterns, limiting their ability to engage latent propulsive capacity in real time. Most of these protocols utilize passive conditions, such as predetermined belt speeds or externally applied resistance, which constrain speed variability and impair real-time adaptability to individual gait dynamics. In addition, many rehabilitation programs focus primarily on improving step or stance symmetry rather than directly enhancing propulsive force, thereby neglecting a critical biomechanical determinant of post-stroke gait recovery. Prolonged exposure to fixed asymmetric belt speeds may also induce maladaptive gait strategies, reinforcing preexisting asymmetries or generating new, inefficient movement patterns [ 17 ]. To overcome the limitations of fixed treadmill parameters, biofeedback-based gait training has been introduced to provide real-time sensory cues aligned with gait mechanics. For instance, one study directly compared the immediate effects of visual, auditory, and combined audiovisual biofeedback on propulsive force generation in both able-bodied individuals and those post-stroke [ 18 ]. The results indicated that auditory biofeedback led to greater increases in peak propulsive force than visual biofeedback among younger adults. However, no significant differences among biofeedback modalities were observed in the post-stroke group or in older adults more generally [ 18 ]. This absence of modality-specific effects in older and neurologically impaired populations may be attributed to age-related sensory decline. In particular, auditory processing often deteriorates earlier and more significantly than visual processing with age, potentially diminishing the relative effectiveness of auditory cues in older adults and individuals post-stroke [ 19 ]. Although the study demonstrated that biofeedback can elicit short-term improvements in propulsive force [ 18 ], its effects were primarily limited to providing augmented visual information without mechanically influencing gait dynamics. Because forward propulsion is largely governed by the timing and magnitude of AGRF during late stance, further enhancement of propulsion may require dynamic modulation of mechanical parameters, such as treadmill belt speed. Integrating such mechanical modulation with visual biofeedback may therefore offer a more comprehensive and effective approach to augment gait propulsion in individuals post-stroke. In response to these limitations, we recently developed the Adaptive Propulsion Enhancement eXperience (APEX) system and conducted a preliminary evaluation of its impact on gait propulsion in individuals post-stroke [ 20 ]. The APEX system utilizes a programmable split-belt treadmill integrated with force plates to detect key gait events (i.e., heel strike and toe-off) and gait cycle in real time. Based on this real-time detection, the system dynamically adjusts belt speed during the late stance phase (i.e., from mid push-off to toe-off) of the affected leg. This adaptive belt speed modulation is intended to prolong ground contact time and facilitate increased voluntary propulsive effort, a mechanism termed the propulsion-facilitating mode. The propulsion-facilitating mode is coupled with intuitive visual biofeedback to help individuals post-stroke focus attention on modulating push-off force. In our previous feasibility study, this combined intervention resulted in a 39% increase in peak propulsive force and a 31% increase in propulsive impulse during the training period, with these improvements retained at 33% and 35%, respectively, during the post-training period in individuals post-stroke [ 20 ]. Despite the promising results demonstrated in the initial validation of the APEX system, the previous study evaluated only a combined intervention consisting of visual biofeedback and adaptive belt speed modulation. As a result, it remains unclear whether visual biofeedback, propulsion-facilitating mode, or their combination was primarily responsible for the observed improvements in propulsive force generation. To clarify the unique and synergistic contributions of each component, it is necessary to assess their independent and combined effects. It may be possible that some components prove redundant and can be removed to reduce system complexity and cost. Otherwise, it is also possible that both visual biofeedback and propulsion-facilitation are essential for augmenting AGRF. To determine the effective and possibly more efficient configuration, we investigated how the motor outcomes depend on the different mode configurations. This ablation study may not only contribute to refining future system by identifying the essential components for the augmentation of AGRF, but also enhance the clinical relevance of the APEX system and support the development of scalable protocols for post-stroke gait rehabilitation. 2. METHODS 2.1. Participants Prior to participant recruitment, sample size estimation was performed based on our previous study [ 20 ], which evaluated the combined effects of visual biofeedback and adaptive propulsion facilitation using the APEX system. That study demonstrated a large effect size (Cohen’s f > 0.8) for the primary kinetic outcome measure (i.e., peak propulsive force), measured across the training periods. Using this effect size, a power analysis was conducted with G*Power 3.1, which indicated that a minimum of ten participants would be required to detect significant within-subject effects with 80% power at a two-sided alpha level of 0.05, assuming a moderate correlation (r = 0.5). To ensure adequate statistical power while accounting for potential participant withdrawal and inter-individual variability, thirteen individuals with chronic hemiparetic stroke (8 females, 5 males; mean age = 60.5 ± 7.6 years) participated in this study. Inclusion criteria were: (1) ≥ 6 months post-stroke, (2) a single, unilateral, supratentorial ischemic stroke confirmed through medical records, (3) observable motor impairment in the affected leg (e.g., reduced push-off force), and (4) ability to ambulate independently for at least 10 meters. Exclusion criteria included: (1) history of cerebellar or brainstem stroke, (2) other neurologic, orthopedic, or cardiometabolic conditions affecting gait, (3) uncontrolled hypertension (systolic > 180 mmHg or diastolic > 110 mmHg), (4) botulinum toxin injection to the affected leg within the past six months, (5) inability to walk on a treadmill for 30 minutes with rest as needed, (6) clinically significant cognitive impairment, and (7) current or suspected pregnancy. All participants provided written informed consent prior to study enrollment. The study protocol was approved by the Institutional Review Board of the University of Houston and conducted in accordance with the Declaration of Helsinki. 2.2. Adaptive Propulsion Enhancement eXperience (APEX) System The APEX system is a real-time gait training platform developed to enhance forward propulsive force generation during gait rehabilitation. While the technical specifications and full system architecture are detailed in our previous publication [ 19 ], this section provides a concise summary relevant to the present study. The APEX system comprises a split-belt instrumented treadmill (Bertec Corporation, Columbus, OH, USA) and custom-developed software, as illustrated in Fig. 1 . The split-belt instrumented treadmill is equipped with dual independent belts, each embedded with force plates that measure three-dimensional ground reaction forces (GRFs) separately for the left and right legs. This hardware configuration allows for real-time measurement of anterior-posterior (APGRF) and vertical ground reaction forces (VGRF). The custom-developed software serves as the central control and data processing framework within the APEX system. It is implemented in Microsoft Visual C + + as a real-time, multi-threaded application, ensuring continuous acquisition and processing of GRF signals and precise control of the treadmill belts. The software samples GRF signals from the two force plates of the split-belt instrumented treadmill at an update rate of 100 Hz and applies a second-order low-pass Butterworth filter with a cutoff frequency of 10 Hz to remove high-frequency noise in the GRF signals. Using these filtered VGRF data, a reliable gait phase detection algorithm identifies key gait events, including heel strike and toe-off, and segments each phase and gait cycle. This algorithm, based on predefined thresholds normalized to participant body weight, enables real-time identification of gait events with high temporal precision. A distinctive feature of the software is its real-time control of the treadmill belts to deliver propulsion-promotion interventions. In particular, the propulsion-facilitating mode of the APEX system involves a gradual reduction of the treadmill belt speed during the late stance phase, specifically from the midpoint of push-off to toe-off. This strategic deceleration is designed to prolong the ground contact time of the affected leg, thereby encouraging participants to generate greater voluntary propulsive forces and enhance self-generated propulsion during gait training. Following toe-off, the belt corresponding to the affected leg is promptly returned to the participant’s preferred walking speed (PWS), ensuring a smooth transition and continuous walking cadence. Meanwhile, the belt corresponding to the unaffected leg is maintained at the PWS throughout each gait cycle to preserve bilateral gait symmetry and natural gait patterns. In addition to the propulsion-facilitating mode, the software provides real-time visual biofeedback on the propulsive force. Specifically, the magnitude of the propulsive force generated by the affected leg is displayed graphically as a dynamic bar graph on an external monitor, with the reference propulsive force of the unaffected leg serving as the target value. When the propulsive force of the affected leg exceeds the reference propulsive force of the unaffected leg, the bar graph turns green, indicating enhanced propulsion relative to the reference. Conversely, when the propulsive force of the affected leg is below the reference threshold, the bar graph turns red, indicating a deficit. This real-time feedback provides participants with precise, moment-to-moment insights into their propulsive performance relative to their unaffected leg, thereby facilitating informed, adaptive modifications to their gait strategy. The software incorporates comprehensive data logging to ensure high-resolution recording of all relevant variables, including filtered GRF signals, gait event detection outputs, and belt speed modulation commands. All data streams are synchronously stored at 100 Hz to facilitate detailed offline data analysis. Furthermore, to address the present study’s aim of comparing the effects of different modalities (i.e., visual biofeedback alone, propulsion-facilitating mode alone, and their combination), the APEX system’s modular software architecture enables the independent or simultaneous activation of these assistive features. As a result, this modular control allows for precise evaluation of the unique and combined impacts of each training modality on gait propulsion. 2.3. Experimental protocols The experimental apparatus comprised the APEX system, a wearable motion capture system (MVN Awinda, Xsens Technologies, NL), and a wireless electromyography (EMG) system (Trigno TMIM, Delsys Inc., Natick, MA, USA). The motion capture system captured lower-limb kinematics using inertial measurement units (IMUs) that were attached to the sternum, pelvis, left and right upper legs, left and right lower legs, and left and right feet. Each sensor was secured with elastic straps, and both the method of attachment and the sensor placement were in accordance with the manufacturer’s instructions. Based on our previous findings [ 20 ], which indicated that training with the APEX system did not significantly affect dorsiflexors such as the tibialis anterior or knee flexors such as the medial hamstrings during the late stance phase (i.e., from mid push-off to toe-off), EMG data collection in the present study was limited to muscles primarily responsible for propulsion. Specifically, EMG sensors were attached to the plantarflexors (medial gastrocnemius (MG) and soleus (SOL)) and knee extensors (vastus medialis (VM) and rectus femoris (RF)) of the affected leg. The placement of IMU and EMG sensors is illustrated in Fig. 2 . Following sensor placement, participants were fitted with a safety harness (Fig. 2 ) and instructed to walk on the split-belt treadmill to determine their preferred walking speed (PWS). A trained researcher gradually increased or decreased the belt speed while asking participants to report a pace that felt natural and comfortable, reflecting their typical overground walking pattern [ 20 ]. After confirming the PWS, participants walked 30 steps at this speed. The average of the peak AGRF of the unaffected leg across these 30 steps was calculated and later used as the reference value for visual biofeedback. To acclimate to the experimental conditions, participants completed three practice trials, each consisting of 15 steps, under the following modalities: visual biofeedback (VB) alone, propulsion-facilitating mode (PF) alone, and combined visual biofeedback with propulsion-facilitating mode (VB + PF). Participants then completed three separate experimental trials, one for each of the aforementioned training conditions (VB, PF, and VB + PF). The order of these trials was randomized and counterbalanced across participants to reduce order and learning effects. A seated rest period of at least 10 minutes was provided between trials to minimize fatigue and prevent carry-over effects. Consistent with the experimental design adopted in our previous study [ 19 ], each trial consisted of three periods: (1) a baseline period of 30 steps without VB, PF, or VB + PF, (2) a training period of 100 steps with the designated modality applied, and (3) a post-training period of 30 steps without any applied modality (i.e., no VB, PF, or VB + PF). During the baseline and post-training periods, treadmill belt speed was held constant at each participant’s PWS. In the VB condition, real-time visual biofeedback on the AGRF of the affected leg was provided through a dynamic bar graph displayed on an external monitor, as illustrated in Fig. 2 . The bar turned green when the AGRF exceeded the predefined target, and red when it did not. The target was defined as the average peak AGRF of the unaffected leg measured during the baseline phase. In the PF condition, the treadmill belt corresponding to the affected leg underwent controlled deceleration during the late stance phase, beginning at mid push-off and continuing until toe-off. This modulation aimed to prolong ground contact time and encourage active generation of propulsive force. No visual biofeedback was presented in this condition. In the VB + PF condition, both visual biofeedback and propulsion-facilitating mode were simultaneously activated during the training phase. This configuration allowed participants to benefit from combined perceptual cues and mechanical facilitation intended to enhance propulsive output. 2.4. Data and statistical analysis All biomechanical and electrophysiological data were processed offline using MATLAB (MathWorks Inc., Natick, MA, USA). The kinetic outcome measure was the peak propulsive force normalized to body weight, generated by the affected leg and defined as the maximum AGRF during the push-off phase of the gait cycle. Kinematic outcome measures included (1) stride length, (2) knee joint angle, and (3) ankle joint angle, all measured from the IMU-based motion capture system. All kinematic data were filtered using a second-order low-pass Butterworth filter with a 10 Hz cutoff frequency to minimize motion artifacts and high-frequency noise. Stride length was calculated as the anteroposterior displacement between consecutive heel strikes of the same foot. Knee and ankle joint angles of the affected leg were computed in the sagittal plane. Sagittal plane peak extension angles of the knee and ankle were analyzed considering their biomechanical relevance to push-off during gait. Raw EMG signals collected from MG, SOL, VM, and RF muscles of the affected leg were band-pass filtered using a fifth-order Butterworth filter with cutoff frequencies of 20 Hz and 300 Hz to isolate the frequency band relevant to muscle activity. The filtered signals were full-wave rectified, and root mean square (RMS) values were computed for each gait cycle to quantify muscle activation. Following the data analysis approach adopted in our previous study [ 19 ], all outcome measures were computed on a step-by-step basis, with each gait cycle defined as the interval from one heel strike to the subsequent heel strike of the same foot. To minimize step-to-step variability and account for intra-individual variability in performance, each outcome measure was normalized to the average of the first ten steps within each trial. Following normalization, values were averaged across each of the three experimental periods (baseline, training, post-training). For EMG analysis, RMS values were also computed specifically during the belt speed modulation period, corresponding to the late stance phase (i.e., from mid push-off to toe-off) of the affected leg to precisely characterize muscle activation during the propulsive portion of stance. For all outcome measures, distributional assumptions were assessed using the Shapiro-Wilk test to inform the selection of appropriate statistical models. For outcome measures satisfying the assumption of normality, a two-way repeated-measures ANOVA (RMANOVA) was performed with modalities (VB, PF, VB + PF) and periods (baseline, training, post-training) as within-subject factors. For non-normally distributed outcome measures, generalized estimating equations (GEE) with an exchangeable working correlation structure were used to model within-subject dependencies. Main effects of modality and period, as well as their interaction (modality × period), were examined. When significant main or interaction effects were observed, pairwise comparisons using the least significant difference method were conducted to examine differences among modalities and periods. All statistical analyses were performed using SPSS Statistics (version 29, IBM Corp., Armonk, NY, USA), with statistical significance defined as P < 0.05. 3. RESULTS 3.1. Demographic information and clinical characteristics of the participants Table I summarizes the demographic and clinical characteristics of the participants. All individuals were in the chronic stage of stroke recovery, ensuring a stable neurological status throughout the study. Their PWS were considerably lower than normative values, consistent with gait impairments commonly observed in individuals post-stroke. All participants completed the full experimental protocol without adverse events or protocol deviations. 3.2. Peak propulsive force and associated gait kinematics of the affected leg Table 2 summarizes the results of statistical analyses examining the effects of modality, period, and their interaction on normalized peak propulsive force, normalized stride length, normalized peak knee extension, and normalized peak ankle extension. Significant main effects of both modality and period were observed for all four outcome measures, as well as significant modality × period interaction effects. Pairwise comparisons within each modality indicated that all four outcome measures (i.e., normalized peak propulsive force, normalized stride length, normalized peak knee extension, and normalized peak ankle extension) were significantly greater during both the training and post-training periods compared to the baseline period (P < 0.0001), regardless of modality. However, no significant differences were found between the training and post-training periods for any of the outcome measures for each modality, indicating sustained effects following training. Figure 3 illustrates multiple pairwise comparisons across the modalities and periods for all kinetic and kinematic outcome measures: normalized peak propulsive force (A), normalized stride length (B), normalized peak knee extension (C), and normalized peak ankle extension (D). For all outcome measures, the combined visual biofeedback with propulsion-facilitating mode (i.e., VB + PF) resulted in the most substantial improvements. Specifically, normalized peak propulsive force increased by 35% (from 1.00 to 1.35) during training and 37% (from 1.00 to 1.37) during post-training, normalized stride length by 36% (from 1.00 to 1.36) and 32% (from 1.00 to 1.32), normalized peak knee extension by 28% (from 1.02 to 1.31) and 27% (from 1.01 to 1.28), and normalized peak ankle extension by 27% (from 1.03 to 1.31) and 26% (from 1.03 to 1.30), respectively, compared to baseline. The propulsion-facilitating mode (i.e., PF) alone led to substantial improvements across all kinetic and kinematic outcome measures. Specifically, normalized peak propulsive force increased from 1.01 at baseline to 1.33 during training and 1.31 during post-training, corresponding to a 32% increase and a 30% increase. Normalized stride length increased from 1.01 to 1.19 during training and 1.21 during post-training, representing an 18% and 20% increase, respectively. Normalized peak knee extension increased from 1.01 to 1.18 during training and 1.17 during post-training, indicating a 17% and 16% increase. Normalized peak ankle extension increased from 0.99 at baseline to 1.18 during training and 1.17 during post-training, corresponding to a 19% increase and a 18% increase. Visual biofeedback (i.e., VB) alone resulted in more modest yet statistically significant improvements. Normalized peak propulsive force increased from 1.00 at baseline to 1.19 during training and 1.24 during post-training, corresponding to a 19% and 24% increase. Normalized stride length increased from 0.99 to 1.10 during training and 1.12 during post-training, demonstrating a 10% and 12% increase, respectively. Normalized peak knee extension increased from 1.01 to 1.08 during training and 1.09 during post-training, indicating an 8% and 9% increase. Normalized peak ankle extension increased from 0.99 at baseline to 1.10 during training and remained unchanged during post-training, representing a consistent 10% increase after training. 3.3. Muscle activity of the affected leg Table 3 summarizes the results of statistical analyses examining the effects of modality, period, and their interaction on normalized RMS EMG activity of the MG, SOL, VM, and RF of the affected leg. Significant main effects of both modality and period were observed for all four muscles, as well as significant modality × period interaction effects. Pairwise comparisons within each modality revealed that muscle activity was significantly greater during both the training and post-training periods compared to the baseline period (P < 0.0001), regardless of modality. However, no significant differences were found between the training and post-training periods in any muscle, indicating that the neuromuscular adaptations induced by training were retained during the post-training. Figure 4 presents the pairwise comparisons of normalized RMS EMG activity across modalities and periods for each muscle. For all muscles, the combined visual biofeedback with propulsion-facilitating mode (i.e., VB + PF) resulted in the most substantial increases. Specifically, normalized MG activity increased by 38% (from 0.96 to 1.32) during training and post-training, normalized SOL by 33% (from 0.97 to 1.29) and 35% (from 0.97 to 1.31), normalized VM by 33% (from 0.97 to 1.29) and 36% (from 0.97 to 1.32), and normalized RF by 36% (from 0.96 to 1.31) and 35% (from 0.96 to 1.30), respectively, compared to baseline. The propulsion-facilitating mode (i.e., PF) alone led to substantial improvements in muscle activation. Specifically, normalized MG activity increased from 0.97 at baseline to 1.20 during training and 1.19 during post-training, corresponding to a 24% and 23% increase. normalized SOL activity increased from 0.96 to 1.20 during training and slightly decreased to 1.19 during post-training (24% and 23% increases, respectively), normalized VM activity increased from 0.97 to 1.22 and 1.23 (26% and 27% increases), and normalized RF activity increased from 0.96 to 1.20 and 1.23 (24% and 27% increases). Visual biofeedback (i.e., VB) alone showed more modest yet statistically significant increases in muscle activity. Specifically, normalized MG activity increased from 0.99 at baseline to 1.10 during training and 1.13 during post-training, corresponding to 11% and 14% increases, respectively. Normalized SOL activity increased from 0.96 to 1.12 during training and remained at 1.12 during post-training (17% increase maintained). Normalized VM activity increased from 0.97 to 1.12 during training and slightly decreased to 1.11 during post-training (15% and 14% increases, respectively), while normalized RF activity increased from 0.97 to 1.14 during training and 1.13 during post-training (18% and 16% increases). 4. DISCUSSION This study examined the independent and combined effects of visual biofeedback (i.e., VB) and the propulsion-facilitating mode (i.e., PF) on gait propulsion and neuromechanical performance in individuals post-stroke. The results confirmed that VB + PF resulted in greater improvements in affected leg’s propulsive force, kinematics (i.e., knee and ankle extension), and EMG activity in the ankle and knee extensor muscles (i.e., medial gastrocnemius, soleus, vastus medialis, and rectus femoris) than either VB or PF alone. These findings not only reinforce the concept of an underutilized propulsive capacity in the affected leg but also highlight the value of simultaneous perceptual and gait phase-specific mechanical modulation (i.e., adaptive belt speed modulation through the PF) for optimizing gait rehabilitation outcomes in post-stroke populations. 4.1. Synergistic and differential effects of visual biofeedback and propulsion-facilitating mode Training with the combined VB and PF resulted in the most substantial enhancements across all outcome measures, including a 35% increase in peak propulsive force, a 36% increase in stride length, a 31% increase in peak knee extension, a 31% increase in peak ankle extension, and approximately a 30% increase in EMG activity of major ankle and knee extensors (Figs. 3 and 4 ). This pattern suggests a synergistic effect between visual-motor engagement and adaptive belt speed modulation during the stance phase. While the PF increased biomechanical demand by prolonging ground contact time through controlled belt deceleration during late stance, VB augmented volitional effort by providing real-time visual feedback, thereby enabling participants to actively adjust their gait patterns. The integration of these complementary mechanisms likely facilitated more effective utilization of the latent propulsion reserve, which has been previously documented in stroke survivors as an underutilized capacity that can be recruited to enhance functional recovery [ 2 , 7 ]. In contrast, training with VB alone resulted in comparatively modest enhancements. Although VB can facilitate motor adaptation by improving attentional focus and reinforcing task-relevant goals [ 21 ], it does not provide the dynamic, spatiotemporally targeted augmentation offered by the PF. The observation that PF alone led to greater enhancements than VB alone suggests that increasing propulsive demands through stance-phase-specific belt speed modulation serves as a critical mechanism for eliciting neuromechanical engagement. This finding is consistent with prior research indicating that propulsion can be enhanced by manipulating gait mechanics using resistance, asymmetry, or incline-based strategies [ 15 , 16 , 22 ]. 4.2. Propulsion-facilitating mode and latent propulsive capacity The findings associated with the PF provide compelling evidence for the existence of a latent propulsive reserve in individuals post-stroke [ 23 ]. In this study, the PF adaptively modulated treadmill belt speed during the late stance phase (i.e., from mid push-off to toe-off) of the affected leg, thereby increasing ground contact time and enhancing biomechanical conditions for effective push-off. Consistent with prior studies demonstrating that prolonging the stance phase facilitates plantarflexor engagement in individuals post-stroke [ 24 , 25 ], the present study found that the PF elicited greater activation in both proximal (e.g., vastus medialis, rectus femoris) and distal (e.g., gastrocnemius, soleus) muscles compared to VB alone (Fig. 4 ). This pattern supports a more comprehensive neuromuscular recruitment strategy during the push-off phase. Importantly, our prior work using the APEX system demonstrated that adaptive belt speed modulation during the late stance phase does not induce destabilizing effects or compensatory activation patterns, as indicated by the absence of increased activity in dorsiflexors or hip abductors during the modulation phase [ 20 ]. Together with the current findings, these results provide critical validation for the safety of the propulsion-targeted strategy and informed the design of the present study, which specifically aimed to leverage the latent propulsive capacity in individuals post-stroke. 4.3. Short-term aftereffects and evidence of motor adaptation Training with all three modalities (i.e., VB, PF, and VB + PF) led to short-term aftereffects, with improvements in overall gait propulsion of the affected leg maintained during the post-training period even after each modality was withdrawn. The retention of improvements in propulsive force, joint kinematics, and muscle activity in the absence of ongoing VB, PF, or their combination indicates that short-term motor adaptation occurred during training. This observation is consistent with previous studies showing that both error-augmented feedback and mechanical perturbation–based training can induce motor learning-like aftereffects in individuals post-stroke [ 18 , 21 , 24 , 26 ]. Notably, the greatest degree of aftereffects was observed in training with VB + PF, suggesting that multimodal interventions that simultaneously target both neural control and task-specific mechanical demands are more likely to induce more persistent short-term modifications in gait strategy. Furthermore, the presence of aftereffects across all modalities indicates that even brief, targeted gait training can promote adaptive changes that persist beyond the immediate training period. Such short-term persistence supports the clinical scalability of concise, phase-specific gait interventions aimed at augmenting propulsion, particularly when multimodal cues are used to concurrently engage cognitive, sensory, and mechanical pathways. 4.4. Clinical implications The clinical implications of our findings can be articulated in three primary aspects. First, we suggested a feasible method for directly targeting propulsion of the affected leg, which remains an often underemphasized yet essential determinant of walking function after stroke [ 2 , 5 ]. Impaired propulsive force in individuals post-stroke has been associated with reduced walking speed, increased metabolic cost, and limited capacity for community ambulation [ 6 ]. In particular, the reported beneficial effects of PF alone suggest the potential of this modality to facilitate the rehabilitation of individuals with limited attentional ability. While training with VB + PF resulted in the most pronounced and sustained gains in propulsive function, training with PF alone also provided promising results. Therefore, for individuals with cognitive or attentional impairments who have difficulty in processing or responding to real-time visual biofeedback, the mechanically driven, phase-specific facilitation provided by PF can still enhance late-stance propulsive output without imposing additional cognitive load. Accordingly, interventions that can safely and effectively augment gait propulsion, such as the VB, PF, and VB + PF protocols demonstrated in this study, may provide a critical element of post-stroke gait rehabilitation strategies that aim to restore functional mobility. Second, the modular architecture of the APEX system enhances its potential for scalable integration into clinical practice. In resource-limited settings, simplified VB can be implemented using low-cost motion capture devices or pressure-sensitive insoles. Similarly, propulsion-facilitating strategies may be approximated through incline walking or resistive harness systems. In clinics equipped with advanced technology, systems such as APEX offer real-time gait phase detection and individualized belt speed modulation, thereby enabling patient-specific training that is temporally and mechanically optimized for the propulsion needs of the affected leg. Lastly, the immediate enhancements in gait propulsion observed following short-duration exposure suggest that propulsion-focused training may be effective even within the constraints of typical clinical session lengths. Integrating short periods of VB + PF walking into conventional rehabilitation protocols has the potential to provide additive therapeutic benefits without significantly disrupting routine clinical workflows. 4.5. Limitations and Future Directions This study has several limitations that necessitate further investigation. First, the sample size was relatively small, and all participants were in the chronic phase of stroke recovery. Future studies should include larger and more heterogeneous cohorts, including individuals in the subacute stage, to better assess the generalizability of the findings. Second, the training was limited to a single-session exposure. Although short-term aftereffects were observed, future research is needed to investigate the impact of multi-session training protocols on cumulative gains, long-term retention, and transferability to overground walking. Third, real-time biofeedback in this study was delivered solely through the visual modality. Given the variability in sensory processing among individuals post-stroke [ 27 ], future research should explore alternative feedback modalities, such as auditory or haptic cues, and investigate adaptive strategies that integrate user preferences to enable personalized feedback delivery. 5. CONCLUSION This study systematically investigated the independent and combined effects of VB and PF on gait propulsion and neuromechanical performance in individuals post-stroke. While each training modality (i.e., VB and PF alone) independently led to significant enhancements in propulsive force, gait kinematics, and EMG activity in the ankle and knee extensor muscles, their combination (i.e., VB + PF) induced the most substantial enhancements. These findings highlight the synergistic advantages of simultaneously engaging perceptual and mechanical pathways to facilitate the recruitment of underutilized propulsive capacity in the affected leg. Importantly, enhancements observed during the training period were retained across all modalities during the post-training period. This finding indicates that short-duration gait training, whether delivered through a single modality (i.e., VB or PF alone) or a multimodal (i.e., VB + PF) approach, can elicit aftereffects that are consistent with motor adaptation. By differentiating the independent and combined contributions of VB and adaptive belt speed modulation (i.e., PF), this study offers a more refined understanding of the mechanisms underlying propulsion enhancement and supports the clinical feasibility of modular and scalable gait training strategies tailored to individual needs. Taken together, these findings provide compelling support for the implementation of real-time, phase-specific, multimodal gait training paradigms that target both neural and biomechanical determinants of gait propulsion. Future research should investigate the long-term efficacy of real-time gait training that integrates both VB and PF, including the optimal training frequency and duration, and assess its generalizability to overground walking as well as its applicability across diverse neurologically impaired populations. Declarations Ethics approval and consent to participate All procedures were approved by the Institutional Review Board of the University of Houston (IRB#: STUDY00004224). Written informed consent was obtained from all participants prior to data collection. Consent for publication Consent for publication was given by all participants. Competing interests The authors declare no competing interests. Funding Research was supported by the Brain Pool Program funded by the Ministry of Science and ICT (MSIT) through the National Research Foundation of Korea under Grant RS-2024-00446461, in part by Korea Health Technology Research and Development Project through Korea Health Industry Development Institute (KHIDI) funded by the Ministry of Health and Welfare under Grant HK23C0071, and in part by the National Research Foundation of Korea Grant funded by Korean Government (MSIT) under Grant RS-2023-00208052. Author Contribution S.H.P., H.P., and B.-C.L. conceived and designed research. B.-C.L. developed and implemented the experimental system. S.H.P., H.P., and B.-C.L. performed experiments. C.L. and B.-C.L. analyzed data and performed statistical analysis. S.H.P., C.L., H.P., J.A., and B.-C.L interpreted the results. C.L., H.P., and B.-C.L prepared figures. S.H.P., C.L., and B.-C.L drafted the manuscript. S.H.P., C.L., H.P., J.A., and B.-C.L edited and revised the manuscript. J.A. and B.-C.L approved the final version of the manuscript. Acknowledgement The authors would like to express their sincere gratitude to all participants for their time and commitment to this study. The authors also gratefully acknowledge Joshua Doan, Yasmeen Elfeki, Ria Kolluru, Joshua Lim, Nhat Nguyen, Maya Palitz, and Adriele Rivera for their invaluable assistance with data collection. Data Availability The data from the current study are available from the corresponding authors on reasonable request. References Feigin VL, et al. World Stroke Organization (WSO): global stroke fact sheet 2022. Int J stroke. 2022;17(1):18–29. Awad LN, et al. These legs were made for propulsion: advancing the diagnosis and treatment of post-stroke propulsion deficits. J Neuroeng Rehabil. 2020;17(1):139. Hsu A-L, Tang P-F, Jan M-H. Analysis of impairments influencing gait velocity and asymmetry of hemiplegic patients after mild to moderate stroke. Arch Phys Med Rehabil. 2003;84(8):1185–93. Michael KM, Allen JK, Macko RF. Reduced ambulatory activity after stroke: the role of balance, gait, and cardiovascular fitness. Arch Phys Med Rehabil. 2005;86(8):1552–6. Bowden MG et al. Anterior-posterior ground reaction forces as a measure of paretic leg contribution in hemiparetic walking. Stroke, 2006. Ohta M, et al. Patterns of change in propulsion force and late braking force in patients with stroke walking at comfortable and fast speeds. Sci Rep. 2024;14(1):22316. Lewek MD, Raiti C, Doty A. The presence of a paretic propulsion reserve during gait in individuals following stroke. Neurorehabilit Neural Repair. 2018;32(12):1011–9. Todhunter-Brown A et al. Physical rehabilitation approaches for the recovery of function and mobility following stroke . Cochrane Database of Systematic Reviews, 2025(2). McEwen SE, et al. Cognitive strategy use to enhance motor skill acquisition post-stroke: a critical review. Brain Injury. 2009;23(4):263–77. Guzik A, Drużbicki M, Wolan-Nieroda A. Assessment of two gait training models: conventional physical therapy and treadmill exercise, in terms of their effectiveness after stroke. Hippokratia. 2018;22(2):51–9. Tyrell CM, Helm E, Reisman DS. Locomotor adaptation is influenced by the interaction between perturbation and baseline asymmetry after stroke. J Biomech. 2015;48(11):2849–57. Reisman DS, et al. Split-belt treadmill adaptation transfers to overground walking in persons poststroke. Neurorehabilit Neural Repair. 2009;23(7):735–44. Reisman DS, et al. Locomotor adaptation on a split-belt treadmill can improve walking symmetry post-stroke. Brain. 2007;130(7):1861–72. Dzewaltowski AC, et al. The effect of split-belt treadmill interventions on step length asymmetry in individuals poststroke: a systematic review with meta-analysis. Neurorehabilit Neural Repair. 2021;35(7):563–75. Moradian N, et al. Effects of backward-directed resistance on propulsive force generation during split-belt treadmill walking in non-impaired individuals. Front Hum Neurosci. 2023;17:1214967. Phadke CP. Immediate effects of a single inclined treadmill walking session on level ground walking in individuals after stroke. Am J Phys Med Rehabil. 2012;91(4):337–45. Betschart M, McFayden BJ, Nadeau S. Lower limb joint moments on the fast belt contribute to a reduction of step length asymmetry over ground after split-belt treadmill training in stroke: A pilot study. Physiotherapy theory and practice; 2020. Liu J, et al. Comparison of the immediate effects of audio, visual, or audiovisual gait biofeedback on propulsive force generation in able-bodied and post-stroke individuals. Appl Psychophysiol Biofeedback. 2020;45(3):211–20. Wang J, Puel J-L. Presbycusis: an update on cochlear mechanisms and therapies. J Clin Med. 2020;9(1):218. Park SH, et al. A novel Adaptive Propulsion Enhancement eXperience (APEX) System: Development and preliminary validation for enhancing gait propulsion in stroke survivors. IEEE Trans Neural Syst Rehabil Eng. 2025;33:1486–96. Genthe K, et al. Effects of real-time gait biofeedback on paretic propulsion and gait biomechanics in individuals post-stroke. Top Stroke Rehabil. 2018;25(3):186–93. Hsiao H, et al. Mechanisms to increase propulsive force for individuals poststroke. J Neuroeng Rehabil. 2015;12(1):40. Alingh JF, et al. Task-specific training for improving propulsion symmetry and gait speed in people in the chronic phase after stroke: a proof-of-concept study. J Neuroeng Rehabil. 2021;18(1):69. Reisman DS, Bastian AJ, Morton SM. Neurophysiologic and Rehabilitation Insights From the Split-Belt and Other Locomotor Adaptation Paradigms. Phys Ther. 2010;90(2):187–95. Shin SY, et al. Relationship between gait quality measures and modular neuromuscular control parameters in chronic post-stroke individuals. J Neuroeng Rehabil. 2021;18(1):58. Park H, et al. Transcutaneous spinal stimulation paired with visual feedback facilitates retention of improved weight transfer toward the affected side in people post-stroke. J Neuroeng Rehabil. 2025;22(1):188. Progress in motor control: neural, computational and dynamic approaches . Advances in Experimental Medicine and Biology. Vol. 782. 2013, New York, NY: Springer New York. Tables Table 1. Demographic information and clinical characteristics of the participants (n=13). Abbreviation: F, female; M, male; PWS, preferred walking speed Metric Sex (F=1, M=0) Age (years) Body mass (kg) Height (cm) Post-injury (years) PWS (m/s) Mean 0.6 60.5 74.3 170.1 7.5 0.4 Standard deviation 0.5 7.6 13.8 7.0 5.6 0.2 Table 2. Results of statistical analysis for normalized peak propulsive force and associated gait kinematics of the affected leg. Abbreviation: M, Modality; P, Period; DF, degrees of freedom Outcome measure Effects DF P value Peak propulsive force M 2 < 0.0001 P 2 < 0.0001 M × P 4 < 0.0001 Stride length M 2 < 0.0001 P 2 < 0.0001 M × P 4 < 0.0001 Peak knee extension M 2 < 0.0001 P 2 < 0.0001 M × P 4 < 0.0001 Peak ankle extension M 2 < 0.0001 P 2 < 0.0001 M × P 4 < 0.0001 Table 3. Results of statistical analysis for normalized muscle activity of the affected leg. Abbreviation: RMS, root mean square; EMG, electromyography; M, Modality; P, Period; DF, degrees of freedom Outcome measure Effects DF P value RMS EMG activity of medial gastrocnemius (MG) M 2 < 0.0001 P 2 < 0.0001 M × P 4 < 0.0001 RMS EMG activity of soleus (SOL) M 2 < 0.0001 P 2 < 0.0001 M × P 4 < 0.0001 RMS EMG activity of vastus medialis (VM) M 2 < 0.0001 P 2 < 0.0001 M × P 4 < 0.0001 RMS EMG activity of rectus femoris (RF) M 2 < 0.0001 P 2 < 0.0001 M × P 4 < 0.0001 Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 12 May, 2026 Reviewers agreed at journal 12 May, 2026 Reviews received at journal 12 Jan, 2026 Reviewers agreed at journal 15 Dec, 2025 Reviewers agreed at journal 08 Dec, 2025 Reviewers invited by journal 08 Dec, 2025 Editor assigned by journal 20 Nov, 2025 Submission checks completed at journal 20 Nov, 2025 First submitted to journal 19 Nov, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8159960","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":557505001,"identity":"accbc062-c1d1-4c1f-b7f7-3fc1abb72ac5","order_by":0,"name":"Seoung Hoon Park","email":"","orcid":"","institution":"University of Houston","correspondingAuthor":false,"prefix":"","firstName":"Seoung","middleName":"Hoon","lastName":"Park","suffix":""},{"id":557505002,"identity":"9a0dd411-dc62-4fad-89a5-dd97a5897a88","order_by":1,"name":"Chihyeong Lee","email":"","orcid":"","institution":"Seoul National University","correspondingAuthor":false,"prefix":"","firstName":"Chihyeong","middleName":"","lastName":"Lee","suffix":""},{"id":557505003,"identity":"fba7b659-d910-40ed-99b2-164f6d2a7e75","order_by":2,"name":"Hyunje Park","email":"","orcid":"","institution":"University of Houston","correspondingAuthor":false,"prefix":"","firstName":"Hyunje","middleName":"","lastName":"Park","suffix":""},{"id":557505004,"identity":"285a88b3-7295-4aa9-9f41-5540a7cda7a0","order_by":3,"name":"Jooeun Ahn","email":"","orcid":"","institution":"Seoul National University","correspondingAuthor":false,"prefix":"","firstName":"Jooeun","middleName":"","lastName":"Ahn","suffix":""},{"id":557505005,"identity":"a001fbeb-fb7f-42cc-abf4-3dfa1bee6dab","order_by":4,"name":"Beom-Chan Lee","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA7ElEQVRIiWNgGAWjYBACA3Yg8YHHBsotAIoAKQm8WpgZGBhnyKQBmcwMYPVEaWHmsTlMghZzZuaDD3hyzsvzS/cf+/DDgCFxOwPzwds8eLRYNrMlG0icuW04c85h5pk9QC07G9iSrfFpMTjMYyZh2HM7weBGMjMDD1DLhgM8ZtL4tfB//5H471yCPVAL4x+wFv5vBLTwsDEc4DmQYCCRzMwMtYUNrxagX4wlG3iSDWfcSDZmljGQMN5wmM3Ycg4eLebszQ8//+Gxk+efkfiY8U2FjeyG480Pb7zBowUdSEDiZxSMglEwCkYBZQAAyypEDDn+w+QAAAAASUVORK5CYII=","orcid":"","institution":"University of Houston","correspondingAuthor":true,"prefix":"","firstName":"Beom-Chan","middleName":"","lastName":"Lee","suffix":""}],"badges":[],"createdAt":"2025-11-20 03:38:18","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8159960/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8159960/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":98423442,"identity":"e5aed901-851b-4688-be32-0e7e5a0eb3d4","added_by":"auto","created_at":"2025-12-17 16:32:13","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":608597,"visible":true,"origin":"","legend":"","description":"","filename":"JNERManuscript.docx","url":"https://assets-eu.researchsquare.com/files/rs-8159960/v1/32b8d743af8ca43a34842dcb.docx"},{"id":97976820,"identity":"5c958243-eb31-42ec-a880-1cfb69878003","added_by":"auto","created_at":"2025-12-11 12:07:32","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":8216,"visible":true,"origin":"","legend":"","description":"","filename":"59a5cbf8396d4179bb2f3c37b663666f.json","url":"https://assets-eu.researchsquare.com/files/rs-8159960/v1/b46d23ce86a0695e11192060.json"},{"id":98424635,"identity":"d6ff0042-71cf-40a5-9cb9-0392e0f0f1ed","added_by":"auto","created_at":"2025-12-17 16:33:36","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":107197,"visible":true,"origin":"","legend":"","description":"","filename":"59a5cbf8396d4179bb2f3c37b663666f1enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8159960/v1/4d78dec083dc028ddde18379.xml"},{"id":97976826,"identity":"1ab4ed8a-88d9-49e2-9d5e-167484572e2f","added_by":"auto","created_at":"2025-12-11 12:07:32","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":30918,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8159960/v1/8d4fbdfc3df6b949d9d2f8eb.png"},{"id":97976822,"identity":"db6296dd-f17f-41a7-8df9-a5ad62ebce7a","added_by":"auto","created_at":"2025-12-11 12:07:32","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":49163,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8159960/v1/77e901dde94c809cee760090.png"},{"id":97976824,"identity":"c963d629-fc54-469e-944b-59fb7bde5fb5","added_by":"auto","created_at":"2025-12-11 12:07:32","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":52713,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8159960/v1/72e16817673d7d59928f1893.png"},{"id":97976829,"identity":"5424516e-df55-4865-b4ae-697758abf28e","added_by":"auto","created_at":"2025-12-11 12:07:32","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":205686,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8159960/v1/1a2088eabff20ece9b88f935.png"},{"id":97976832,"identity":"aad0c44c-fd0e-4e36-91e0-2c8acaa2b35d","added_by":"auto","created_at":"2025-12-11 12:07:33","extension":"xml","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":105382,"visible":true,"origin":"","legend":"","description":"","filename":"59a5cbf8396d4179bb2f3c37b663666f1structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8159960/v1/a625ac5cc35cd375df7b7540.xml"},{"id":97976833,"identity":"56cd6ee6-34e1-449a-88ca-1ff69e8bf02b","added_by":"auto","created_at":"2025-12-11 12:07:33","extension":"html","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":115186,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8159960/v1/fb4c22e726f55e5513cf0340.html"},{"id":98424610,"identity":"5d6cd81a-5de4-4e6d-8ad9-779ba4084881","added_by":"auto","created_at":"2025-12-17 16:33:33","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":94815,"visible":true,"origin":"","legend":"\u003cp\u003eOverview of the APEX system enabling real-time gait training through visual biofeedback and phase-specific adaptive belt speed modulation during the late stance phase.\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8159960/v1/9200b23ec10f96a3f0aff813.png"},{"id":98424536,"identity":"f4c35cfc-89f5-43d1-9b79-2f8c21cf313b","added_by":"auto","created_at":"2025-12-17 16:33:26","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":164941,"visible":true,"origin":"","legend":"\u003cp\u003eExperimental setup for real-time gait training using the APEX system.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8159960/v1/c7b855c48df69daf5101c6b0.png"},{"id":98423591,"identity":"05e18598-bc2f-40dc-861a-4d810ae73968","added_by":"auto","created_at":"2025-12-17 16:32:23","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":102978,"visible":true,"origin":"","legend":"\u003cp\u003eAverage peak propulsive force and associated gait kinematics of the affected leg across modalities and periods. \u003cstrong\u003eA.\u003c/strong\u003e Normalized peak propulsive force. \u003cstrong\u003eB.\u003c/strong\u003e Normalized stride length. \u003cstrong\u003eC.\u003c/strong\u003e Normalized peak knee extension. \u003cstrong\u003eD.\u003c/strong\u003e Normalized peak ankle extension.Each bar represents the group average across all participants, and error bars indicate the standard deviation. *, **, and *** indicate p \u0026lt; 0.05, p \u0026lt; 0.01, and p \u0026lt; 0.0001, respectively. Abbreviation: B, baseline; T, training; P-T, post-training; VB, visual biofeedback; PF, propulsion-facilitating\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8159960/v1/6c893ee07a386cb87dcb218c.png"},{"id":98423552,"identity":"b038af7d-6180-4b0a-9654-09760e7dfa07","added_by":"auto","created_at":"2025-12-17 16:32:21","extension":"jpeg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":456837,"visible":true,"origin":"","legend":"\u003cp\u003eAverage RMS EMG activity of the affected leg across modalities and periods. \u003cstrong\u003eA.\u003c/strong\u003e Medial gastrocnemius (MG). \u003cstrong\u003eB.\u003c/strong\u003e Soleus (SOL). \u003cstrong\u003eC.\u003c/strong\u003e Vastus medialis (VM). \u003cstrong\u003eD.\u003c/strong\u003e Rectus femoris (RF).Each bar represents the group average across all participants, and error bars indicate the standard deviation. *, **, and *** indicate p \u0026lt; 0.05, p \u0026lt; 0.01, and p \u0026lt; 0.0001, respectively. Abbreviation: B, baseline; T, training; P-T, post-training; VB, visual biofeedback; PF, propulsion-facilitating\u003c/p\u003e","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8159960/v1/e01fd523f8d293c3afa628d5.jpeg"},{"id":98622995,"identity":"4dce1934-2bb5-4b6e-8d3d-bacf3964286f","added_by":"auto","created_at":"2025-12-19 17:03:56","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1669060,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8159960/v1/33b4599d-c2ef-478d-85c1-d5b6659bdab1.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Independent and synergistic effects of visual biofeedback and phase-specific belt deceleration on affected-leg propulsion in post-stroke split-belt treadmill training","fulltext":[{"header":"1. BACKGROUND","content":"\u003cp\u003eStroke remains one of the most prevalent causes of gait dysfunction, frequently leading to substantial impairments in ambulatory capacity and overall functional mobility [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. These impairments are commonly characterized by reduced walking speed, compromised postural control, gait asymmetry, and increased metabolic cost, all of which are closely associated with increased risks of falls, musculoskeletal strain, early-onset fatigue, and a significant decline in quality of life [\u003cspan additionalcitationids=\"CR3\" citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Among the various biomechanical deficits observed in individuals post-stroke, impairments in push-off mechanics during the late stance phase (i.e., from mid push-off to toe-off) are particularly consequential. These impairments reduce the generation of anterior ground reaction force (AGRF), a key determinant of forward propulsion that facilitates body progression and supports symmetrical step advancement [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eGiven these impairments, restoring AGRF, particularly from the affected leg, has emerged as a central therapeutic target in gait rehabilitation. Indeed, achieving a stable and efficient gait in individuals post-stroke requires sufficient and coordinated generation of AGRF from both legs, with particular emphasis on the affected leg. Inadequate generation of AGRF in the affected leg leads to reduced propulsive output, asymmetric step transitions, and increased reliance on compensatory mechanisms, ultimately compromising gait stability and performance. Previous studies have reported that individuals post-stroke generate only about 16 to 49% of the total propulsive force using the affected leg [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. More recently, data from a cohort of 100 stroke survivors demonstrated a significant positive association between increased propulsion from the affected leg and improved walking speed, underscoring the clinical relevance of this parameter [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Notably, emerging evidence suggests that individuals post-stroke retain a latent capacity to generate greater propulsive force from the affected leg [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. This phenomenon, referred to as the affected leg propulsion reserve, reflects an underutilized capacity that may be recruited to facilitate functional recovery. Taken together, these findings emphasize that restoring forward propulsion, particularly through targeted enhancement of AGRF in the affected leg, represents a critical therapeutic objective in post-stroke gait rehabilitation.\u003c/p\u003e\u003cp\u003eTo address propulsion-related deficits, various conventional rehabilitation approaches have been implemented. These include manual therapeutic interventions such as therapeutic exercise, functional task-specific training, and the motor relearning program, which aim to restore normal movement patterns through therapist-guided facilitation and practice. Traditional interventions primarily focus on improving muscular strength, flexibility, balance, and motor control through repetitive, structured, and cognitively engaging exercises [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. However, such conventional manual therapeutic approaches (e.g., neuro-developmental handling and facilitation-based gait training) provide limited direct enhancement of the anterior ground reaction force (AGRF), a key biomechanical determinant of forward propulsion [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Rather than explicitly strengthening the propulsive capacity of the affected leg, these approaches emphasize gait symmetry, joint mobility, and postural alignment. Consequently, improvements in walking speed or step length are often attributable to compensatory mechanisms rather than restoration of forward propulsion mechanics [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eThe limited efficacy of manual interventions in enhancing AGRF has motivated the exploration of more targeted, task-specific approaches. Other studies have specifically targeted the enhancement of forward propulsion by encouraging increased generation of anterior ground reaction force (AGRF) through treadmill-based gait training. One commonly studied method utilizes split-belt treadmill protocols that apply asymmetrical belt speeds (for example, a 2:1 speed ratio), with the goal of reducing gait asymmetries in individuals post-stroke [\u003cspan additionalcitationids=\"CR12\" citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. A recent meta-analysis reported that this type of training can improve interlimb symmetry, and in some instances, the benefits are maintained beyond the intervention period [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Additionally, other studies have examined training conditions that increase the biomechanical demands of propulsion, including walking against backward-directed resistance [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e] or walking on an inclined surface [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eThese interventions, while biomechanically demanding, often lack adaptive responsiveness to individual gait patterns, limiting their ability to engage latent propulsive capacity in real time. Most of these protocols utilize passive conditions, such as predetermined belt speeds or externally applied resistance, which constrain speed variability and impair real-time adaptability to individual gait dynamics. In addition, many rehabilitation programs focus primarily on improving step or stance symmetry rather than directly enhancing propulsive force, thereby neglecting a critical biomechanical determinant of post-stroke gait recovery. Prolonged exposure to fixed asymmetric belt speeds may also induce maladaptive gait strategies, reinforcing preexisting asymmetries or generating new, inefficient movement patterns [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eTo overcome the limitations of fixed treadmill parameters, biofeedback-based gait training has been introduced to provide real-time sensory cues aligned with gait mechanics. For instance, one study directly compared the immediate effects of visual, auditory, and combined audiovisual biofeedback on propulsive force generation in both able-bodied individuals and those post-stroke [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. The results indicated that auditory biofeedback led to greater increases in peak propulsive force than visual biofeedback among younger adults. However, no significant differences among biofeedback modalities were observed in the post-stroke group or in older adults more generally [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. This absence of modality-specific effects in older and neurologically impaired populations may be attributed to age-related sensory decline. In particular, auditory processing often deteriorates earlier and more significantly than visual processing with age, potentially diminishing the relative effectiveness of auditory cues in older adults and individuals post-stroke [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. Although the study demonstrated that biofeedback can elicit short-term improvements in propulsive force [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e], its effects were primarily limited to providing augmented visual information without mechanically influencing gait dynamics. Because forward propulsion is largely governed by the timing and magnitude of AGRF during late stance, further enhancement of propulsion may require dynamic modulation of mechanical parameters, such as treadmill belt speed. Integrating such mechanical modulation with visual biofeedback may therefore offer a more comprehensive and effective approach to augment gait propulsion in individuals post-stroke.\u003c/p\u003e\u003cp\u003eIn response to these limitations, we recently developed the Adaptive Propulsion Enhancement eXperience (APEX) system and conducted a preliminary evaluation of its impact on gait propulsion in individuals post-stroke [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. The APEX system utilizes a programmable split-belt treadmill integrated with force plates to detect key gait events (i.e., heel strike and toe-off) and gait cycle in real time. Based on this real-time detection, the system dynamically adjusts belt speed during the late stance phase (i.e., from mid push-off to toe-off) of the affected leg. This adaptive belt speed modulation is intended to prolong ground contact time and facilitate increased voluntary propulsive effort, a mechanism termed the propulsion-facilitating mode. The propulsion-facilitating mode is coupled with intuitive visual biofeedback to help individuals post-stroke focus attention on modulating push-off force. In our previous feasibility study, this combined intervention resulted in a 39% increase in peak propulsive force and a 31% increase in propulsive impulse during the training period, with these improvements retained at 33% and 35%, respectively, during the post-training period in individuals post-stroke [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eDespite the promising results demonstrated in the initial validation of the APEX system, the previous study evaluated only a combined intervention consisting of visual biofeedback and adaptive belt speed modulation. As a result, it remains unclear whether visual biofeedback, propulsion-facilitating mode, or their combination was primarily responsible for the observed improvements in propulsive force generation. To clarify the unique and synergistic contributions of each component, it is necessary to assess their independent and combined effects. It may be possible that some components prove redundant and can be removed to reduce system complexity and cost. Otherwise, it is also possible that both visual biofeedback and propulsion-facilitation are essential for augmenting AGRF. To determine the effective and possibly more efficient configuration, we investigated how the motor outcomes depend on the different mode configurations. This ablation study may not only contribute to refining future system by identifying the essential components for the augmentation of AGRF, but also enhance the clinical relevance of the APEX system and support the development of scalable protocols for post-stroke gait rehabilitation.\u003c/p\u003e"},{"header":"2. METHODS","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e2.1. Participants\u003c/h2\u003e\u003cp\u003ePrior to participant recruitment, sample size estimation was performed based on our previous study [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e], which evaluated the combined effects of visual biofeedback and adaptive propulsion facilitation using the APEX system. That study demonstrated a large effect size (Cohen\u0026rsquo;s f\u0026thinsp;\u0026gt;\u0026thinsp;0.8) for the primary kinetic outcome measure (i.e., peak propulsive force), measured across the training periods. Using this effect size, a power analysis was conducted with G*Power 3.1, which indicated that a minimum of ten participants would be required to detect significant within-subject effects with 80% power at a two-sided alpha level of 0.05, assuming a moderate correlation (r\u0026thinsp;=\u0026thinsp;0.5).\u003c/p\u003e\u003cp\u003eTo ensure adequate statistical power while accounting for potential participant withdrawal and inter-individual variability, thirteen individuals with chronic hemiparetic stroke (8 females, 5 males; mean age\u0026thinsp;=\u0026thinsp;60.5\u0026thinsp;\u0026plusmn;\u0026thinsp;7.6 years) participated in this study. Inclusion criteria were: (1)\u0026thinsp;\u0026ge;\u0026thinsp;6 months post-stroke, (2) a single, unilateral, supratentorial ischemic stroke confirmed through medical records, (3) observable motor impairment in the affected leg (e.g., reduced push-off force), and (4) ability to ambulate independently for at least 10 meters. Exclusion criteria included: (1) history of cerebellar or brainstem stroke, (2) other neurologic, orthopedic, or cardiometabolic conditions affecting gait, (3) uncontrolled hypertension (systolic\u0026thinsp;\u0026gt;\u0026thinsp;180 mmHg or diastolic\u0026thinsp;\u0026gt;\u0026thinsp;110 mmHg), (4) botulinum toxin injection to the affected leg within the past six months, (5) inability to walk on a treadmill for 30 minutes with rest as needed, (6) clinically significant cognitive impairment, and (7) current or suspected pregnancy.\u003c/p\u003e\u003cp\u003e All participants provided written informed consent prior to study enrollment. The study protocol was approved by the Institutional Review Board of the University of Houston and conducted in accordance with the Declaration of Helsinki.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003e2.2. Adaptive Propulsion Enhancement eXperience (APEX) System\u003c/h2\u003e\u003cp\u003eThe APEX system is a real-time gait training platform developed to enhance forward propulsive force generation during gait rehabilitation. While the technical specifications and full system architecture are detailed in our previous publication [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e], this section provides a concise summary relevant to the present study. The APEX system comprises a split-belt instrumented treadmill (Bertec Corporation, Columbus, OH, USA) and custom-developed software, as illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\u003cp\u003eThe split-belt instrumented treadmill is equipped with dual independent belts, each embedded with force plates that measure three-dimensional ground reaction forces (GRFs) separately for the left and right legs. This hardware configuration allows for real-time measurement of anterior-posterior (APGRF) and vertical ground reaction forces (VGRF). The custom-developed software serves as the central control and data processing framework within the APEX system. It is implemented in Microsoft Visual C\u0026thinsp;+\u0026thinsp;+\u0026thinsp;as a real-time, multi-threaded application, ensuring continuous acquisition and processing of GRF signals and precise control of the treadmill belts.\u003c/p\u003e\u003cp\u003eThe software samples GRF signals from the two force plates of the split-belt instrumented treadmill at an update rate of 100 Hz and applies a second-order low-pass Butterworth filter with a cutoff frequency of 10 Hz to remove high-frequency noise in the GRF signals. Using these filtered VGRF data, a reliable gait phase detection algorithm identifies key gait events, including heel strike and toe-off, and segments each phase and gait cycle. This algorithm, based on predefined thresholds normalized to participant body weight, enables real-time identification of gait events with high temporal precision.\u003c/p\u003e\u003cp\u003eA distinctive feature of the software is its real-time control of the treadmill belts to deliver propulsion-promotion interventions. In particular, the propulsion-facilitating mode of the APEX system involves a gradual reduction of the treadmill belt speed during the late stance phase, specifically from the midpoint of push-off to toe-off. This strategic deceleration is designed to prolong the ground contact time of the affected leg, thereby encouraging participants to generate greater voluntary propulsive forces and enhance self-generated propulsion during gait training. Following toe-off, the belt corresponding to the affected leg is promptly returned to the participant\u0026rsquo;s preferred walking speed (PWS), ensuring a smooth transition and continuous walking cadence. Meanwhile, the belt corresponding to the unaffected leg is maintained at the PWS throughout each gait cycle to preserve bilateral gait symmetry and natural gait patterns.\u003c/p\u003e\u003cp\u003eIn addition to the propulsion-facilitating mode, the software provides real-time visual biofeedback on the propulsive force. Specifically, the magnitude of the propulsive force generated by the affected leg is displayed graphically as a dynamic bar graph on an external monitor, with the reference propulsive force of the unaffected leg serving as the target value. When the propulsive force of the affected leg exceeds the reference propulsive force of the unaffected leg, the bar graph turns green, indicating enhanced propulsion relative to the reference. Conversely, when the propulsive force of the affected leg is below the reference threshold, the bar graph turns red, indicating a deficit. This real-time feedback provides participants with precise, moment-to-moment insights into their propulsive performance relative to their unaffected leg, thereby facilitating informed, adaptive modifications to their gait strategy.\u003c/p\u003e\u003cp\u003eThe software incorporates comprehensive data logging to ensure high-resolution recording of all relevant variables, including filtered GRF signals, gait event detection outputs, and belt speed modulation commands. All data streams are synchronously stored at 100 Hz to facilitate detailed offline data analysis.\u003c/p\u003e\u003cp\u003eFurthermore, to address the present study\u0026rsquo;s aim of comparing the effects of different modalities (i.e., visual biofeedback alone, propulsion-facilitating mode alone, and their combination), the APEX system\u0026rsquo;s modular software architecture enables the independent or simultaneous activation of these assistive features. As a result, this modular control allows for precise evaluation of the unique and combined impacts of each training modality on gait propulsion.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003e2.3. Experimental protocols\u003c/h2\u003e\u003cp\u003eThe experimental apparatus comprised the APEX system, a wearable motion capture system (MVN Awinda, Xsens Technologies, NL), and a wireless electromyography (EMG) system (Trigno TMIM, Delsys Inc., Natick, MA, USA). The motion capture system captured lower-limb kinematics using inertial measurement units (IMUs) that were attached to the sternum, pelvis, left and right upper legs, left and right lower legs, and left and right feet. Each sensor was secured with elastic straps, and both the method of attachment and the sensor placement were in accordance with the manufacturer\u0026rsquo;s instructions.\u003c/p\u003e\u003cp\u003eBased on our previous findings [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e], which indicated that training with the APEX system did not significantly affect dorsiflexors such as the tibialis anterior or knee flexors such as the medial hamstrings during the late stance phase (i.e., from mid push-off to toe-off), EMG data collection in the present study was limited to muscles primarily responsible for propulsion. Specifically, EMG sensors were attached to the plantarflexors (medial gastrocnemius (MG) and soleus (SOL)) and knee extensors (vastus medialis (VM) and rectus femoris (RF)) of the affected leg. The placement of IMU and EMG sensors is illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e\u003cp\u003eFollowing sensor placement, participants were fitted with a safety harness (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e) and instructed to walk on the split-belt treadmill to determine their preferred walking speed (PWS). A trained researcher gradually increased or decreased the belt speed while asking participants to report a pace that felt natural and comfortable, reflecting their typical overground walking pattern [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. After confirming the PWS, participants walked 30 steps at this speed. The average of the peak AGRF of the unaffected leg across these 30 steps was calculated and later used as the reference value for visual biofeedback.\u003c/p\u003e\u003cp\u003eTo acclimate to the experimental conditions, participants completed three practice trials, each consisting of 15 steps, under the following modalities: visual biofeedback (VB) alone, propulsion-facilitating mode (PF) alone, and combined visual biofeedback with propulsion-facilitating mode (VB\u0026thinsp;+\u0026thinsp;PF). Participants then completed three separate experimental trials, one for each of the aforementioned training conditions (VB, PF, and VB\u0026thinsp;+\u0026thinsp;PF). The order of these trials was randomized and counterbalanced across participants to reduce order and learning effects. A seated rest period of at least 10 minutes was provided between trials to minimize fatigue and prevent carry-over effects.\u003c/p\u003e\u003cp\u003eConsistent with the experimental design adopted in our previous study [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e], each trial consisted of three periods: (1) a baseline period of 30 steps without VB, PF, or VB\u0026thinsp;+\u0026thinsp;PF, (2) a training period of 100 steps with the designated modality applied, and (3) a post-training period of 30 steps without any applied modality (i.e., no VB, PF, or VB\u0026thinsp;+\u0026thinsp;PF). During the baseline and post-training periods, treadmill belt speed was held constant at each participant\u0026rsquo;s PWS.\u003c/p\u003e\u003cp\u003eIn the VB condition, real-time visual biofeedback on the AGRF of the affected leg was provided through a dynamic bar graph displayed on an external monitor, as illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. The bar turned green when the AGRF exceeded the predefined target, and red when it did not. The target was defined as the average peak AGRF of the unaffected leg measured during the baseline phase.\u003c/p\u003e\u003cp\u003eIn the PF condition, the treadmill belt corresponding to the affected leg underwent controlled deceleration during the late stance phase, beginning at mid push-off and continuing until toe-off. This modulation aimed to prolong ground contact time and encourage active generation of propulsive force. No visual biofeedback was presented in this condition.\u003c/p\u003e\u003cp\u003eIn the VB\u0026thinsp;+\u0026thinsp;PF condition, both visual biofeedback and propulsion-facilitating mode were simultaneously activated during the training phase. This configuration allowed participants to benefit from combined perceptual cues and mechanical facilitation intended to enhance propulsive output.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e\u003cb\u003e2.4. Data and statistical analysis\u003c/b\u003e\u003c/h2\u003e\u003cp\u003eAll biomechanical and electrophysiological data were processed offline using MATLAB (MathWorks Inc., Natick, MA, USA). The kinetic outcome measure was the peak propulsive force normalized to body weight, generated by the affected leg and defined as the maximum AGRF during the push-off phase of the gait cycle.\u003c/p\u003e\u003cp\u003eKinematic outcome measures included (1) stride length, (2) knee joint angle, and (3) ankle joint angle, all measured from the IMU-based motion capture system. All kinematic data were filtered using a second-order low-pass Butterworth filter with a 10 Hz cutoff frequency to minimize motion artifacts and high-frequency noise. Stride length was calculated as the anteroposterior displacement between consecutive heel strikes of the same foot. Knee and ankle joint angles of the affected leg were computed in the sagittal plane. Sagittal plane peak extension angles of the knee and ankle were analyzed considering their biomechanical relevance to push-off during gait.\u003c/p\u003e\u003cp\u003eRaw EMG signals collected from MG, SOL, VM, and RF muscles of the affected leg were band-pass filtered using a fifth-order Butterworth filter with cutoff frequencies of 20 Hz and 300 Hz to isolate the frequency band relevant to muscle activity. The filtered signals were full-wave rectified, and root mean square (RMS) values were computed for each gait cycle to quantify muscle activation.\u003c/p\u003e\u003cp\u003eFollowing the data analysis approach adopted in our previous study [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e], all outcome measures were computed on a step-by-step basis, with each gait cycle defined as the interval from one heel strike to the subsequent heel strike of the same foot. To minimize step-to-step variability and account for intra-individual variability in performance, each outcome measure was normalized to the average of the first ten steps within each trial. Following normalization, values were averaged across each of the three experimental periods (baseline, training, post-training). For EMG analysis, RMS values were also computed specifically during the belt speed modulation period, corresponding to the late stance phase (i.e., from mid push-off to toe-off) of the affected leg to precisely characterize muscle activation during the propulsive portion of stance.\u003c/p\u003e\u003cp\u003eFor all outcome measures, distributional assumptions were assessed using the Shapiro-Wilk test to inform the selection of appropriate statistical models. For outcome measures satisfying the assumption of normality, a two-way repeated-measures ANOVA (RMANOVA) was performed with modalities (VB, PF, VB\u0026thinsp;+\u0026thinsp;PF) and periods (baseline, training, post-training) as within-subject factors. For non-normally distributed outcome measures, generalized estimating equations (GEE) with an exchangeable working correlation structure were used to model within-subject dependencies.\u003c/p\u003e\u003cp\u003eMain effects of modality and period, as well as their interaction (modality \u0026times; period), were examined. When significant main or interaction effects were observed, pairwise comparisons using the least significant difference method were conducted to examine differences among modalities and periods. All statistical analyses were performed using SPSS Statistics (version 29, IBM Corp., Armonk, NY, USA), with statistical significance defined as P\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/p\u003e\u003c/div\u003e"},{"header":"3. RESULTS","content":"\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003e3.1. Demographic information and clinical characteristics of the participants\u003c/h2\u003e\u003cp\u003eTable I summarizes the demographic and clinical characteristics of the participants. All individuals were in the chronic stage of stroke recovery, ensuring a stable neurological status throughout the study. Their PWS were considerably lower than normative values, consistent with gait impairments commonly observed in individuals post-stroke. All participants completed the full experimental protocol without adverse events or protocol deviations.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\u003ch2\u003e3.2. Peak propulsive force and associated gait kinematics of the affected leg\u003c/h2\u003e\u003cp\u003eTable\u0026nbsp;2 summarizes the results of statistical analyses examining the effects of modality, period, and their interaction on normalized peak propulsive force, normalized stride length, normalized peak knee extension, and normalized peak ankle extension. Significant main effects of both modality and period were observed for all four outcome measures, as well as significant modality \u0026times; period interaction effects.\u003c/p\u003e\u003cp\u003ePairwise comparisons within each modality indicated that all four outcome measures (i.e., normalized peak propulsive force, normalized stride length, normalized peak knee extension, and normalized peak ankle extension) were significantly greater during both the training and post-training periods compared to the baseline period (P\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), regardless of modality. However, no significant differences were found between the training and post-training periods for any of the outcome measures for each modality, indicating sustained effects following training.\u003c/p\u003e\u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e illustrates multiple pairwise comparisons across the modalities and periods for all kinetic and kinematic outcome measures: normalized peak propulsive force (A), normalized stride length (B), normalized peak knee extension (C), and normalized peak ankle extension (D). For all outcome measures, the combined visual biofeedback with propulsion-facilitating mode (i.e., VB\u0026thinsp;+\u0026thinsp;PF) resulted in the most substantial improvements. Specifically, normalized peak propulsive force increased by 35% (from 1.00 to 1.35) during training and 37% (from 1.00 to 1.37) during post-training, normalized stride length by 36% (from 1.00 to 1.36) and 32% (from 1.00 to 1.32), normalized peak knee extension by 28% (from 1.02 to 1.31) and 27% (from 1.01 to 1.28), and normalized peak ankle extension by 27% (from 1.03 to 1.31) and 26% (from 1.03 to 1.30), respectively, compared to baseline.\u003c/p\u003e\u003cp\u003eThe propulsion-facilitating mode (i.e., PF) alone led to substantial improvements across all kinetic and kinematic outcome measures. Specifically, normalized peak propulsive force increased from 1.01 at baseline to 1.33 during training and 1.31 during post-training, corresponding to a 32% increase and a 30% increase. Normalized stride length increased from 1.01 to 1.19 during training and 1.21 during post-training, representing an 18% and 20% increase, respectively. Normalized peak knee extension increased from 1.01 to 1.18 during training and 1.17 during post-training, indicating a 17% and 16% increase. Normalized peak ankle extension increased from 0.99 at baseline to 1.18 during training and 1.17 during post-training, corresponding to a 19% increase and a 18% increase.\u003c/p\u003e\u003cp\u003eVisual biofeedback (i.e., VB) alone resulted in more modest yet statistically significant improvements. Normalized peak propulsive force increased from 1.00 at baseline to 1.19 during training and 1.24 during post-training, corresponding to a 19% and 24% increase. Normalized stride length increased from 0.99 to 1.10 during training and 1.12 during post-training, demonstrating a 10% and 12% increase, respectively. Normalized peak knee extension increased from 1.01 to 1.08 during training and 1.09 during post-training, indicating an 8% and 9% increase. Normalized peak ankle extension increased from 0.99 at baseline to 1.10 during training and remained unchanged during post-training, representing a consistent 10% increase after training.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\u003ch2\u003e3.3. Muscle activity of the affected leg\u003c/h2\u003e\u003cp\u003eTable\u0026nbsp;3 summarizes the results of statistical analyses examining the effects of modality, period, and their interaction on normalized RMS EMG activity of the MG, SOL, VM, and RF of the affected leg. Significant main effects of both modality and period were observed for all four muscles, as well as significant modality \u0026times; period interaction effects.\u003c/p\u003e\u003cp\u003ePairwise comparisons within each modality revealed that muscle activity was significantly greater during both the training and post-training periods compared to the baseline period (P\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), regardless of modality. However, no significant differences were found between the training and post-training periods in any muscle, indicating that the neuromuscular adaptations induced by training were retained during the post-training.\u003c/p\u003e\u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e presents the pairwise comparisons of normalized RMS EMG activity across modalities and periods for each muscle. For all muscles, the combined visual biofeedback with propulsion-facilitating mode (i.e., VB\u0026thinsp;+\u0026thinsp;PF) resulted in the most substantial increases. Specifically, normalized MG activity increased by 38% (from 0.96 to 1.32) during training and post-training, normalized SOL by 33% (from 0.97 to 1.29) and 35% (from 0.97 to 1.31), normalized VM by 33% (from 0.97 to 1.29) and 36% (from 0.97 to 1.32), and normalized RF by 36% (from 0.96 to 1.31) and 35% (from 0.96 to 1.30), respectively, compared to baseline.\u003c/p\u003e\u003cp\u003eThe propulsion-facilitating mode (i.e., PF) alone led to substantial improvements in muscle activation. Specifically, normalized MG activity increased from 0.97 at baseline to 1.20 during training and 1.19 during post-training, corresponding to a 24% and 23% increase. normalized SOL activity increased from 0.96 to 1.20 during training and slightly decreased to 1.19 during post-training (24% and 23% increases, respectively), normalized VM activity increased from 0.97 to 1.22 and 1.23 (26% and 27% increases), and normalized RF activity increased from 0.96 to 1.20 and 1.23 (24% and 27% increases).\u003c/p\u003e\u003cp\u003eVisual biofeedback (i.e., VB) alone showed more modest yet statistically significant increases in muscle activity. Specifically, normalized MG activity increased from 0.99 at baseline to 1.10 during training and 1.13 during post-training, corresponding to 11% and 14% increases, respectively. Normalized SOL activity increased from 0.96 to 1.12 during training and remained at 1.12 during post-training (17% increase maintained). Normalized VM activity increased from 0.97 to 1.12 during training and slightly decreased to 1.11 during post-training (15% and 14% increases, respectively), while normalized RF activity increased from 0.97 to 1.14 during training and 1.13 during post-training (18% and 16% increases).\u003c/p\u003e\u003c/div\u003e"},{"header":"4. DISCUSSION","content":"\u003cp\u003eThis study examined the independent and combined effects of visual biofeedback (i.e., VB) and the propulsion-facilitating mode (i.e., PF) on gait propulsion and neuromechanical performance in individuals post-stroke. The results confirmed that VB\u0026thinsp;+\u0026thinsp;PF resulted in greater improvements in affected leg\u0026rsquo;s propulsive force, kinematics (i.e., knee and ankle extension), and EMG activity in the ankle and knee extensor muscles (i.e., medial gastrocnemius, soleus, vastus medialis, and rectus femoris) than either VB or PF alone. These findings not only reinforce the concept of an underutilized propulsive capacity in the affected leg but also highlight the value of simultaneous perceptual and gait phase-specific mechanical modulation (i.e., adaptive belt speed modulation through the PF) for optimizing gait rehabilitation outcomes in post-stroke populations.\u003c/p\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003e4.1. Synergistic and differential effects of visual biofeedback and propulsion-facilitating mode\u003c/h2\u003e\u003cp\u003eTraining with the combined VB and PF resulted in the most substantial enhancements across all outcome measures, including a 35% increase in peak propulsive force, a 36% increase in stride length, a 31% increase in peak knee extension, a 31% increase in peak ankle extension, and approximately a 30% increase in EMG activity of major ankle and knee extensors (Figs.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e and \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). This pattern suggests a synergistic effect between visual-motor engagement and adaptive belt speed modulation during the stance phase. While the PF increased biomechanical demand by prolonging ground contact time through controlled belt deceleration during late stance, VB augmented volitional effort by providing real-time visual feedback, thereby enabling participants to actively adjust their gait patterns. The integration of these complementary mechanisms likely facilitated more effective utilization of the latent propulsion reserve, which has been previously documented in stroke survivors as an underutilized capacity that can be recruited to enhance functional recovery [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eIn contrast, training with VB alone resulted in comparatively modest enhancements. Although VB can facilitate motor adaptation by improving attentional focus and reinforcing task-relevant goals [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e], it does not provide the dynamic, spatiotemporally targeted augmentation offered by the PF. The observation that PF alone led to greater enhancements than VB alone suggests that increasing propulsive demands through stance-phase-specific belt speed modulation serves as a critical mechanism for eliciting neuromechanical engagement. This finding is consistent with prior research indicating that propulsion can be enhanced by manipulating gait mechanics using resistance, asymmetry, or incline-based strategies [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e].\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003e4.2. Propulsion-facilitating mode and latent propulsive capacity\u003c/h2\u003e\u003cp\u003eThe findings associated with the PF provide compelling evidence for the existence of a latent propulsive reserve in individuals post-stroke [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. In this study, the PF adaptively modulated treadmill belt speed during the late stance phase (i.e., from mid push-off to toe-off) of the affected leg, thereby increasing ground contact time and enhancing biomechanical conditions for effective push-off.\u003c/p\u003e\u003cp\u003eConsistent with prior studies demonstrating that prolonging the stance phase facilitates plantarflexor engagement in individuals post-stroke [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e], the present study found that the PF elicited greater activation in both proximal (e.g., vastus medialis, rectus femoris) and distal (e.g., gastrocnemius, soleus) muscles compared to VB alone (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). This pattern supports a more comprehensive neuromuscular recruitment strategy during the push-off phase.\u003c/p\u003e\u003cp\u003eImportantly, our prior work using the APEX system demonstrated that adaptive belt speed modulation during the late stance phase does not induce destabilizing effects or compensatory activation patterns, as indicated by the absence of increased activity in dorsiflexors or hip abductors during the modulation phase [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. Together with the current findings, these results provide critical validation for the safety of the propulsion-targeted strategy and informed the design of the present study, which specifically aimed to leverage the latent propulsive capacity in individuals post-stroke.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\u003ch2\u003e4.3. Short-term aftereffects and evidence of motor adaptation\u003c/h2\u003e\u003cp\u003eTraining with all three modalities (i.e., VB, PF, and VB\u0026thinsp;+\u0026thinsp;PF) led to short-term aftereffects, with improvements in overall gait propulsion of the affected leg maintained during the post-training period even after each modality was withdrawn. The retention of improvements in propulsive force, joint kinematics, and muscle activity in the absence of ongoing VB, PF, or their combination indicates that short-term motor adaptation occurred during training. This observation is consistent with previous studies showing that both error-augmented feedback and mechanical perturbation\u0026ndash;based training can induce motor learning-like aftereffects in individuals post-stroke [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eNotably, the greatest degree of aftereffects was observed in training with VB\u0026thinsp;+\u0026thinsp;PF, suggesting that multimodal interventions that simultaneously target both neural control and task-specific mechanical demands are more likely to induce more persistent short-term modifications in gait strategy. Furthermore, the presence of aftereffects across all modalities indicates that even brief, targeted gait training can promote adaptive changes that persist beyond the immediate training period. Such short-term persistence supports the clinical scalability of concise, phase-specific gait interventions aimed at augmenting propulsion, particularly when multimodal cues are used to concurrently engage cognitive, sensory, and mechanical pathways.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\u003ch2\u003e4.4. Clinical implications\u003c/h2\u003e\u003cp\u003eThe clinical implications of our findings can be articulated in three primary aspects. First, we suggested a feasible method for directly targeting propulsion of the affected leg, which remains an often underemphasized yet essential determinant of walking function after stroke [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Impaired propulsive force in individuals post-stroke has been associated with reduced walking speed, increased metabolic cost, and limited capacity for community ambulation [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. In particular, the reported beneficial effects of PF alone suggest the potential of this modality to facilitate the rehabilitation of individuals with limited attentional ability. While training with VB\u0026thinsp;+\u0026thinsp;PF resulted in the most pronounced and sustained gains in propulsive function, training with PF alone also provided promising results. Therefore, for individuals with cognitive or attentional impairments who have difficulty in processing or responding to real-time visual biofeedback, the mechanically driven, phase-specific facilitation provided by PF can still enhance late-stance propulsive output without imposing additional cognitive load. Accordingly, interventions that can safely and effectively augment gait propulsion, such as the VB, PF, and VB\u0026thinsp;+\u0026thinsp;PF protocols demonstrated in this study, may provide a critical element of post-stroke gait rehabilitation strategies that aim to restore functional mobility.\u003c/p\u003e\u003cp\u003eSecond, the modular architecture of the APEX system enhances its potential for scalable integration into clinical practice. In resource-limited settings, simplified VB can be implemented using low-cost motion capture devices or pressure-sensitive insoles. Similarly, propulsion-facilitating strategies may be approximated through incline walking or resistive harness systems. In clinics equipped with advanced technology, systems such as APEX offer real-time gait phase detection and individualized belt speed modulation, thereby enabling patient-specific training that is temporally and mechanically optimized for the propulsion needs of the affected leg.\u003c/p\u003e\u003cp\u003eLastly, the immediate enhancements in gait propulsion observed following short-duration exposure suggest that propulsion-focused training may be effective even within the constraints of typical clinical session lengths. Integrating short periods of VB\u0026thinsp;+\u0026thinsp;PF walking into conventional rehabilitation protocols has the potential to provide additive therapeutic benefits without significantly disrupting routine clinical workflows.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e\u003ch2\u003e4.5. Limitations and Future Directions\u003c/h2\u003e\u003cp\u003eThis study has several limitations that necessitate further investigation. First, the sample size was relatively small, and all participants were in the chronic phase of stroke recovery. Future studies should include larger and more heterogeneous cohorts, including individuals in the subacute stage, to better assess the generalizability of the findings.\u003c/p\u003e\u003cp\u003eSecond, the training was limited to a single-session exposure. Although short-term aftereffects were observed, future research is needed to investigate the impact of multi-session training protocols on cumulative gains, long-term retention, and transferability to overground walking.\u003c/p\u003e\u003cp\u003eThird, real-time biofeedback in this study was delivered solely through the visual modality. Given the variability in sensory processing among individuals post-stroke [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e], future research should explore alternative feedback modalities, such as auditory or haptic cues, and investigate adaptive strategies that integrate user preferences to enable personalized feedback delivery.\u003c/p\u003e\u003c/div\u003e"},{"header":"5. CONCLUSION","content":"\u003cp\u003eThis study systematically investigated the independent and combined effects of VB and PF on gait propulsion and neuromechanical performance in individuals post-stroke. While each training modality (i.e., VB and PF alone) independently led to significant enhancements in propulsive force, gait kinematics, and EMG activity in the ankle and knee extensor muscles, their combination (i.e., VB\u0026thinsp;+\u0026thinsp;PF) induced the most substantial enhancements. These findings highlight the synergistic advantages of simultaneously engaging perceptual and mechanical pathways to facilitate the recruitment of underutilized propulsive capacity in the affected leg.\u003c/p\u003e\u003cp\u003eImportantly, enhancements observed during the training period were retained across all modalities during the post-training period. This finding indicates that short-duration gait training, whether delivered through a single modality (i.e., VB or PF alone) or a multimodal (i.e., VB\u0026thinsp;+\u0026thinsp;PF) approach, can elicit aftereffects that are consistent with motor adaptation. By differentiating the independent and combined contributions of VB and adaptive belt speed modulation (i.e., PF), this study offers a more refined understanding of the mechanisms underlying propulsion enhancement and supports the clinical feasibility of modular and scalable gait training strategies tailored to individual needs.\u003c/p\u003e\u003cp\u003eTaken together, these findings provide compelling support for the implementation of real-time, phase-specific, multimodal gait training paradigms that target both neural and biomechanical determinants of gait propulsion. Future research should investigate the long-term efficacy of real-time gait training that integrates both VB and PF, including the optimal training frequency and duration, and assess its generalizability to overground walking as well as its applicability across diverse neurologically impaired populations.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003cp\u003eAll procedures were approved by the Institutional Review Board of the University of Houston (IRB#: STUDY00004224). Written informed consent was obtained from all participants prior to data collection.\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003cp\u003eConsent for publication was given by all participants.\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003ch2\u003eCompeting interests\u003c/h2\u003e\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\u003c/p\u003e\u003ch2\u003eFunding\u003c/h2\u003e\u003cp\u003eResearch was supported by the Brain Pool Program funded by the Ministry of Science and ICT (MSIT) through the National Research Foundation of Korea under Grant RS-2024-00446461, in part by Korea Health Technology Research and Development Project through Korea Health Industry Development Institute (KHIDI) funded by the Ministry of Health and Welfare under Grant HK23C0071, and in part by the National Research Foundation of Korea Grant funded by Korean Government (MSIT) under Grant RS-2023-00208052.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eS.H.P., H.P., and B.-C.L. conceived and designed research. B.-C.L. developed and implemented the experimental system. S.H.P., H.P., and B.-C.L. performed experiments. C.L. and B.-C.L. analyzed data and performed statistical analysis. S.H.P., C.L., H.P., J.A., and B.-C.L interpreted the results. C.L., H.P., and B.-C.L prepared figures. S.H.P., C.L., and B.-C.L drafted the manuscript. S.H.P., C.L., H.P., J.A., and B.-C.L edited and revised the manuscript. J.A. and B.-C.L approved the final version of the manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eThe authors would like to express their sincere gratitude to all participants for their time and commitment to this study. The authors also gratefully acknowledge Joshua Doan, Yasmeen Elfeki, Ria Kolluru, Joshua Lim, Nhat Nguyen, Maya Palitz, and Adriele Rivera for their invaluable assistance with data collection.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe data from the current study are available from the corresponding authors on reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eFeigin VL, et al. World Stroke Organization (WSO): global stroke fact sheet 2022. Int J stroke. 2022;17(1):18\u0026ndash;29.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAwad LN, et al. These legs were made for propulsion: advancing the diagnosis and treatment of post-stroke propulsion deficits. J Neuroeng Rehabil. 2020;17(1):139.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHsu A-L, Tang P-F, Jan M-H. Analysis of impairments influencing gait velocity and asymmetry of hemiplegic patients after mild to moderate stroke. Arch Phys Med Rehabil. 2003;84(8):1185\u0026ndash;93.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMichael KM, Allen JK, Macko RF. Reduced ambulatory activity after stroke: the role of balance, gait, and cardiovascular fitness. Arch Phys Med Rehabil. 2005;86(8):1552\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBowden MG et al. Anterior-posterior ground reaction forces as a measure of paretic leg contribution in hemiparetic walking. Stroke, 2006.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eOhta M, et al. Patterns of change in propulsion force and late braking force in patients with stroke walking at comfortable and fast speeds. Sci Rep. 2024;14(1):22316.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLewek MD, Raiti C, Doty A. The presence of a paretic propulsion reserve during gait in individuals following stroke. Neurorehabilit Neural Repair. 2018;32(12):1011\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTodhunter-Brown A et al. \u003cem\u003ePhysical rehabilitation approaches for the recovery of function and mobility following stroke\u003c/em\u003e. Cochrane Database of Systematic Reviews, 2025(2).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMcEwen SE, et al. Cognitive strategy use to enhance motor skill acquisition post-stroke: a critical review. Brain Injury. 2009;23(4):263\u0026ndash;77.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGuzik A, Drużbicki M, Wolan-Nieroda A. Assessment of two gait training models: conventional physical therapy and treadmill exercise, in terms of their effectiveness after stroke. Hippokratia. 2018;22(2):51\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTyrell CM, Helm E, Reisman DS. Locomotor adaptation is influenced by the interaction between perturbation and baseline asymmetry after stroke. J Biomech. 2015;48(11):2849\u0026ndash;57.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eReisman DS, et al. Split-belt treadmill adaptation transfers to overground walking in persons poststroke. Neurorehabilit Neural Repair. 2009;23(7):735\u0026ndash;44.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eReisman DS, et al. Locomotor adaptation on a split-belt treadmill can improve walking symmetry post-stroke. Brain. 2007;130(7):1861\u0026ndash;72.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDzewaltowski AC, et al. The effect of split-belt treadmill interventions on step length asymmetry in individuals poststroke: a systematic review with meta-analysis. Neurorehabilit Neural Repair. 2021;35(7):563\u0026ndash;75.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMoradian N, et al. Effects of backward-directed resistance on propulsive force generation during split-belt treadmill walking in non-impaired individuals. Front Hum Neurosci. 2023;17:1214967.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePhadke CP. Immediate effects of a single inclined treadmill walking session on level ground walking in individuals after stroke. Am J Phys Med Rehabil. 2012;91(4):337\u0026ndash;45.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBetschart M, McFayden BJ, Nadeau S. Lower limb joint moments on the fast belt contribute to a reduction of step length asymmetry over ground after split-belt treadmill training in stroke: A pilot study. Physiotherapy theory and practice; 2020.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLiu J, et al. Comparison of the immediate effects of audio, visual, or audiovisual gait biofeedback on propulsive force generation in able-bodied and post-stroke individuals. Appl Psychophysiol Biofeedback. 2020;45(3):211\u0026ndash;20.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang J, Puel J-L. Presbycusis: an update on cochlear mechanisms and therapies. J Clin Med. 2020;9(1):218.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePark SH, et al. A novel Adaptive Propulsion Enhancement eXperience (APEX) System: Development and preliminary validation for enhancing gait propulsion in stroke survivors. IEEE Trans Neural Syst Rehabil Eng. 2025;33:1486\u0026ndash;96.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGenthe K, et al. Effects of real-time gait biofeedback on paretic propulsion and gait biomechanics in individuals post-stroke. Top Stroke Rehabil. 2018;25(3):186\u0026ndash;93.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHsiao H, et al. Mechanisms to increase propulsive force for individuals poststroke. J Neuroeng Rehabil. 2015;12(1):40.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAlingh JF, et al. Task-specific training for improving propulsion symmetry and gait speed in people in the chronic phase after stroke: a proof-of-concept study. J Neuroeng Rehabil. 2021;18(1):69.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eReisman DS, Bastian AJ, Morton SM. Neurophysiologic and Rehabilitation Insights From the Split-Belt and Other Locomotor Adaptation Paradigms. Phys Ther. 2010;90(2):187\u0026ndash;95.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eShin SY, et al. Relationship between gait quality measures and modular neuromuscular control parameters in chronic post-stroke individuals. J Neuroeng Rehabil. 2021;18(1):58.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePark H, et al. Transcutaneous spinal stimulation paired with visual feedback facilitates retention of improved weight transfer toward the affected side in people post-stroke. J Neuroeng Rehabil. 2025;22(1):188.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003e\u003cem\u003eProgress in motor control: neural, computational and dynamic approaches\u003c/em\u003e. Advances in Experimental Medicine and Biology. Vol. 782. 2013, New York, NY: Springer New York.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003e\u003cstrong\u003eTable 1.\u003c/strong\u003e Demographic information and clinical characteristics of the participants (n=13). Abbreviation: F, female; M, male; PWS, preferred walking speed\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"770\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 22.0493%;\"\u003e\n \u003cp\u003eMetric\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12.4514%;\"\u003e\n \u003cp\u003eSex\u003c/p\u003e\n \u003cp\u003e(F=1, M=0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.489%;\"\u003e\n \u003cp\u003eAge\u003c/p\u003e\n \u003cp\u003e(years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15.0454%;\"\u003e\n \u003cp\u003eBody mass\u003c/p\u003e\n \u003cp\u003e(kg)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12.3217%;\"\u003e\n \u003cp\u003eHeight\u003c/p\u003e\n \u003cp\u003e(cm)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12.3217%;\"\u003e\n \u003cp\u003ePost-injury\u003c/p\u003e\n \u003cp\u003e(years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12.3217%;\"\u003e\n \u003cp\u003ePWS\u003c/p\u003e\n \u003cp\u003e(m/s)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 22.0493%;\"\u003e\n \u003cp\u003eMean\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12.4514%;\"\u003e\n \u003cp\u003e0.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.489%;\"\u003e\n \u003cp\u003e60.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15.0454%;\"\u003e\n \u003cp\u003e74.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12.3217%;\"\u003e\n \u003cp\u003e170.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12.3217%;\"\u003e\n \u003cp\u003e7.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12.3217%;\"\u003e\n \u003cp\u003e0.4\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 22.0493%;\"\u003e\n \u003cp\u003eStandard deviation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12.4514%;\"\u003e\n \u003cp\u003e0.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.489%;\"\u003e\n \u003cp\u003e7.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15.0454%;\"\u003e\n \u003cp\u003e13.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12.3217%;\"\u003e\n \u003cp\u003e7.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12.3217%;\"\u003e\n \u003cp\u003e5.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12.3217%;\"\u003e\n \u003cp\u003e0.2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2.\u0026nbsp;\u003c/strong\u003eResults of statistical analysis for normalized peak propulsive force and associated gait kinematics of the affected leg. Abbreviation: M, Modality; P, Period; DF, degrees of freedom\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"631\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 179px;\"\u003e\n \u003cp\u003eOutcome measure\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 228px;\"\u003e\n \u003cp\u003eEffects\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003eDF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003eP value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\" style=\"width: 179px;\"\u003e\n \u003cp\u003ePeak propulsive force\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 228px;\"\u003e\n \u003cp\u003eM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 228px;\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 228px;\"\u003e\n \u003cp\u003eM \u0026times; P\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\" style=\"width: 179px;\"\u003e\n \u003cp\u003eStride length\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 228px;\"\u003e\n \u003cp\u003eM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 228px;\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 228px;\"\u003e\n \u003cp\u003eM \u0026times; P\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\" style=\"width: 179px;\"\u003e\n \u003cp\u003ePeak knee extension\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 228px;\"\u003e\n \u003cp\u003eM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 228px;\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 228px;\"\u003e\n \u003cp\u003eM \u0026times; P\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\" style=\"width: 179px;\"\u003e\n \u003cp\u003ePeak ankle extension\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 228px;\"\u003e\n \u003cp\u003eM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 228px;\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 228px;\"\u003e\n \u003cp\u003eM \u0026times; P\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003eTable 3.\u0026nbsp;\u003c/strong\u003eResults of statistical analysis for normalized muscle activity of the affected leg. Abbreviation: RMS, root mean square; EMG, electromyography; M, Modality; P, Period; DF, degrees of freedom\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"631\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 372px;\"\u003e\n \u003cp\u003eOutcome measure\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003eEffects\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003eDF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003eP value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\" style=\"width: 372px;\"\u003e\n \u003cp\u003eRMS EMG activity of medial gastrocnemius (MG)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003eM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003eM \u0026times; P\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\" style=\"width: 372px;\"\u003e\n \u003cp\u003eRMS EMG activity of soleus (SOL)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003eM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003eM \u0026times; P\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\" style=\"width: 372px;\"\u003e\n \u003cp\u003eRMS EMG activity of vastus medialis (VM)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003eM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003eM \u0026times; P\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\" style=\"width: 372px;\"\u003e\n \u003cp\u003eRMS EMG activity of rectus femoris (RF)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003eM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003eM \u0026times; P\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e\u0026lt; 0.0001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"journal-of-neuroengineering-and-rehabilitation","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jner","sideBox":"Learn more about [Journal of NeuroEngineering and Rehabilitation](http://jneuroengrehab.biomedcentral.com/)","snPcode":"12984","submissionUrl":"https://submission.nature.com/new-submission/12984/3","title":"Journal of NeuroEngineering and Rehabilitation","twitterHandle":"@BioMedCentral","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"post-stroke gait, gait propulsion, adaptive belt speed modulation, visual biofeedback, split-belt treadmill, motor learning","lastPublishedDoi":"10.21203/rs.3.rs-8159960/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8159960/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePost-stroke walking deficits are closely associated with impaired forward propulsion from the affected leg. We developed a real-time training system using an instrumented split-belt treadmill that provides phase-specific adaptive belt speed modulation (propulsion-facilitating mode, PF) during late stance and visual biofeedback (VB) of the affected leg’s propulsive (anterior ground reaction) force. This study investigated the effects of VB, PF, and VB + PF on gait propulsion, kinematics, and muscle activity in individuals post-stroke.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods.\u003c/strong\u003e Thirteen adults with chronic hemiparetic stroke completed three randomized, counterbalanced treadmill trials: VB alone, PF alone, and VB + PF. Each trial consisted of the baseline (30 steps, without intervention), the training (100 steps, assigned modality), and the post-training (30 steps, without intervention) periods. All outcome measures were obtained from the affected leg, including peak propulsive force, stride length, peak knee and ankle extension, and electromyographic activity of the major ankle and knee extensors (i.e., medial gastrocnemius, soleus, vastus medialis, and rectus femoris).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults.\u003c/strong\u003e Significant main effects of modality, period, and their interaction were observed for all outcome measures (P \u0026lt; 0.0001). During the training period, VB + PF resulted in the greatest improvements across all outcome measures (i.e., peak propulsive force, stride length, peak knee and ankle extension, and electromyographic activity in the ankle and knee extensor muscles) compared with either VB or PF alone. PF alone demonstrated moderate improvements, whereas VB alone showed smaller yet statistically significant improvements. Regardless of modality, the improvements achieved during the training period were retained during the post-training period.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion.\u003c/strong\u003e In individuals post-stroke, combining real-time visual biofeedback with phase-specific adaptive belt speed modulation resulted in the greatest and most substantial improvements in affected-leg propulsion, lower-limb kinematics, and electromyographic activity in the ankle and knee extensor muscles. These findings demonstrate the effectiveness of multimodal, adaptive treadmill training in engaging the affected-leg propulsion reserve and inform the development of scalable rehabilitation protocols that integrate perceptual feedback with task-specific mechanical facilitation for stroke gait rehabilitation.\u003c/p\u003e","manuscriptTitle":"Independent and synergistic effects of visual biofeedback and phase-specific belt deceleration on affected-leg propulsion in post-stroke split-belt treadmill training","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-11 12:07:24","doi":"10.21203/rs.3.rs-8159960/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2026-05-12T20:30:22+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"159109379808929891077726563534992226696","date":"2026-05-12T13:29:48+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-01-12T11:10:15+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"61771343713755438528218926254272458858","date":"2025-12-15T21:31:50+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"148034570358305645385979957743938321191","date":"2025-12-08T18:13:25+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-12-08T17:47:25+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-11-20T12:38:05+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-11-20T12:37:23+00:00","index":"","fulltext":""},{"type":"submitted","content":"Journal of NeuroEngineering and Rehabilitation","date":"2025-11-20T03:31:17+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"journal-of-neuroengineering-and-rehabilitation","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jner","sideBox":"Learn more about [Journal of NeuroEngineering and Rehabilitation](http://jneuroengrehab.biomedcentral.com/)","snPcode":"12984","submissionUrl":"https://submission.nature.com/new-submission/12984/3","title":"Journal of NeuroEngineering and Rehabilitation","twitterHandle":"@BioMedCentral","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"ef7006f0-aded-4937-993a-b475138442e7","owner":[],"postedDate":"December 11th, 2025","published":true,"recentEditorialEvents":[{"type":"editorInvitedReview","content":"","date":"2026-05-12T20:30:22+00:00","index":36,"fulltext":""},{"type":"reviewerAgreed","content":"159109379808929891077726563534992226696","date":"2026-05-12T13:29:48+00:00","index":35,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2025-12-11T12:07:24+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-11 12:07:24","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8159960","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8159960","identity":"rs-8159960","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}