Association of Rhythmic Auditory Stimulation Combined with Body Weight Support Treadmill Training with Gait Symmetry and Balance Function in Patients with Post-Stroke Hemiplegia: A Retrospective Study

Abstract

Objective:

To investigate the association of rhythmic auditory stimulation (RAS) combined with body weight support treadmill training (BWSTT) with gait symmetry and balance function in patients with post-stroke hemiplegia.

Methods:

This retrospective study included 110 patients with post-stroke hemiplegia admitted between January 2023 and January 2025. According to the treatment received under non-randomized allocation, patients were divided into the BWSTT-alone group (n = 55) and the RAS + BWSTT group (n = 55). Gait symmetry, balance function (Berg balance scale [BBS] and Fugl–Meyer assessment [FMA-LE], and walking ability (10-m walk test [10MWT] and 6-min walk test [6MWT] were compared between the two groups before and after treatment. Adverse events were recorded.

Results:

After 8 weeks of treatment, the RAS + BWSTT group showed significantly higher step length symmetry index, cadence symmetry index, and double support phase time symmetry index compared with the BWSTT-alone group (P < 0.05). BBS scores, FMA-LE scores, 10MWT speed, and 6MWT distance improved significantly in both groups, with the RAS + BWSTT group showing significantly greater improvement (P < 0.05). However, the between-group differences for BBS (2.26 points) and FMA-LE (2.83 points) did not reach the established minimal clinically important difference (MCID) thresholds. No significant difference in adverse event incidence was observed between the groups (P > 0.05).

Conclusion:

RAS combined with BWSTT was associated with significant improvements in gait symmetry and walking ability compared with BWSTT alone. For balance and lower limb motor function, although between-group differences were statistically significant, they did not meet MCID thresholds; therefore, their clinical significance requires further investigation. This combined approach may provide a reference for developing walking rehabilitation programs for patients with post-stroke hemiplegia.

KEY MESSAGES

• (1)Rhythmic auditory stimulation combined with overhead body weight support treadmill training was associated with greater improvements in gait symmetry, balance function, and walking ability than treadmill training alone in patients with post-stroke hemiplegia.
• (2)The combined therapy demonstrated a favorable safety profile, with no significant increase in adverse events compared with treadmill training alone.
• (3)This combined rehabilitation strategy may address deficits in insufficient weight-bearing, abnormal gait rhythm, and poor symmetry in patients with post-stroke hemiplegia, offering an evidence-based reference for optimizing rehabilitation protocols.

INTRODUCTION

Stroke is the leading cause of mortality and disability among adults in China. According to statistical data, the annual incidence of new stroke cases in China exceeds 2 million, and over half of surviving patients may experience varying degrees of functional impairment, with hemiplegia being the most prevalent manifestation.[1,2] Abnormal gait symmetry and impaired balance function constitute core functional deficits in patients with post-stroke hemiplegia. These impairments not only directly restrict independent ambulation capacity but also affect activities of daily living and social participation, thereby increasing the medical and caregiving burden on both families and society as a whole.[3,4]

Among conventional clinical rehabilitation treatments, body weight support treadmill training (BWSTT) is a commonly employed modality for gait rehabilitation in patients with post-stroke hemiplegia. This approach reduces lower-limb weight-bearing load through body weight support, thereby providing a safe environment for gait training. However, its efficacy in improving gait rhythm and symmetry remains limited when applied alone.[5] Rhythmic auditory stimulation (RAS) is a neurologic music therapy based on neural entrainment and rhythm synchronization. It uses a constant auditory rhythm to guide and regulate stepping movements, thereby potentially contributing to the stability and symmetry of the gait cycle. Nevertheless, the clinical effectiveness of RAS alone may be constrained by patients’ limited lower-limb weight-bearing capacity and impaired balance function.[6]

The present study employed a retrospective analytical design to explore the association between RAS synchronized with BWSTT and gait symmetry and balance function in patients with post-stroke hemiplegia. The findings are intended to generate hypotheses for future prospective trials.

MATERIALS AND METHODS

Study Participants

This retrospective study was based on de-identified clinical data of 138 patients with post-stroke hemiplegia treated at Changxing County Hospital of Traditional Chinese Medicine between January 2023 and January 2025. The ethics committee waived the requirement for informed consent because of the retrospective nature of the study and the use of anonymized data. All procedures complied with the ethical principles outlined in the Declaration of Helsinki.[7] This retrospective study was approved by the Ethics Committee of Changxing County Hospital of Traditional Chinese Medicine (Ethics Approval No. 2026-018-01).

Among the 138 patients, 18 were excluded due to preexisting balance dysfunction or gait impairment, 6 were excluded due to severe visual or auditory impairment, and 4 were excluded due to receiving similar treatment protocols during the same period. Ultimately, 110 eligible patients were enrolled in the final analysis.

Inclusion criteria were as follows: (1) Diagnosis of stroke (either ischemic or hemorrhagic) confirmed by cranial CT or MRI. For ischemic stroke, the diagnosis was in accordance with the Chinese Guidelines for the Diagnosis and Treatment of Acute Ischemic Stroke 2023.[8] (2) First-ever stroke onset, with a disease duration of 1–6 months; (3) Brunnstrom recovery stage III–V of the hemiplegic lower limb; (4) Medical Research Council muscle strength grade ≥3 in the affected lower limb. Grade 3 indicates the ability to perform movement against gravity through the full range of motion, which has been used as an inclusion criterion in previous stroke rehabilitation studies to ensure adequate lower-limb strength for treadmill training; and (5) Mini-mental state examination (MMSE) score ≥24, indicating the ability to cooperate with rehabilitation training and related assessments.[9] The cutoff score of 24 is widely accepted for indicating normal cognitive function in stroke populations, with scores below 24 suggesting the presence of cognitive impairment.[10]

Exclusion criteria were as follows: (1) Concomitant severe functional failure of vital organs (e.g., heart, liver, or kidney) or presence of malignant tumors; (2) Severe visual or auditory impairment precluding the reception of auditory stimulation; (3) Concomitant lower-limb fractures or severe joint contracture deformities; (4) History of balance dysfunction or gait impairment; (5) Participation in other similar treatment protocols during the study period. According to the treatment received, 110 eligible patients with post-stroke hemiplegia were allocated into two groups: the BWSTT-alone group and the RAS + BWSTT group, with 55 patients in each group.

Treatment Method

Patients in both groups received routine pharmacological treatment in the neurology department and conventional rehabilitation training. Conventional rehabilitation training encompassed neuromuscular facilitation techniques, joint range of motion exercises, lower-limb muscle-strengthening exercises, static and dynamic balance training, and activities of daily living training. Training sessions were conducted once daily, with each session lasting 40 min, 5 days per week, for a continuous duration of 8 weeks.

BWSTT-Alone Group

In addition to routine pharmacological treatment and conventional rehabilitation training, patients in this group received BWSTT.

Patients performed walking exercises using the RLMT701 overhead BWSTT system (Guangzhou Renlai Rehabilitation Equipment Manufacturing Co., Ltd., Guangzhou, China). During each training session, patients wore a full-body safety harness and performed walking exercises on a straight treadmill. The initial walking speed was set at 0.5 m/s and was progressively adjusted based on the patient’s gait tolerance and functional improvement. The initial body weight support was set at 10%–30% of the patient’s body weight, adjusted to a level at which the patient could stand with full weight bearing on the lower limbs without notable discomfort. The level of body weight support was reevaluated every 2 weeks and gradually reduced until complete withdrawal of support was achieved. Each training session lasted 30 min and was conducted 5 days per week for a continuous period of 8 weeks.

RAS + BWSTT Group

In addition to the BWSTT protocol, patients in this group received concurrent RAS. Auditory cues with a constant beat frequency were generated using an SQ-70 metronome (Seiko Instruments Inc., Tokyo, Japan) and delivered binaurally via circumaural headphones (Audio-Technica ATH-M20x, frequency response of 15–20,000 Hz, impedance 47 Ω) at a volume that was clearly audible without causing discomfort (estimated 60–70 dB SPL; no calibration meter was used due to the retrospective design). The initial stimulus frequency was set at 90% of the patient’s preferred comfortable cadence, as measured prior to treatment. Hearing safety was ensured by verbal confirmation that no discomfort was experienced during each session, and no adverse auditory events were reported. The beat frequency was adjusted every 2 weeks based on observed improvements in the patient’s gait pattern, with gradual progression toward the normal cadence range of healthy adults. Throughout the training sessions, the same rehabilitation therapist guided the patient to synchronize the timing, frequency, and length of each step with the presented auditory rhythm, ensuring that each footfall precisely coincided with the auditory cue. The total duration, frequency, and treatment period of the training sessions were identical to those of the BWSTT-alone group, with continuous treatment provided over 8 weeks.

Outcome Measures

Data for all outcome measures were extracted from the hospital electronic medical record system and rehabilitation assessment archives at both baseline (pretreatment) and post-treatment (after 8 weeks of treatment) time points. The primary outcome was the step length symmetry index, which was selected a priori as the most direct measure of gait symmetry. Secondary outcome measures included cadence symmetry index, double support phase time symmetry index, BBS score, FMA-LE score, 10-m walk test (10MWT) speed, 6-min walk test (6MWT) distance, and adverse event incidence. Multiple comparison correction was applied using the Bonferroni method with a domain-wise significance threshold as described in the Statistical Analysis section. All 110 enrolled patients had complete medical records covering the 8-week treatment period and post‑treatment assessments. No missing outcome data were identified in the extracted datasets.

Gait Symmetry Parameters

Two-dimensional gait video recordings were acquired by a rehabilitation therapist during the patient’s normal walking. Camera placement: the sagittal plane, at a distance of 3 m, a height of 1 m, and a sampling rate of 30 Hz. No skin markers were used; step length and cadence were derived from visual inspection of foot contact using commercial software (Ningdong Medical Technology, version 2.0). The software has been internally validated against three-dimensional motion capture data (r > 0.85, unpublished data). The video data were subsequently transmitted to Ningdong Medical Technology (Ningbo, China) for gait parameter analysis. The gait analysis software (version 2.0) automatically calculated spatiotemporal gait parameters based on two-dimensional video input. To ensure objectivity, gait analysis was performed by technicians who were blinded to the patients’ group allocation. Patient identifiers were removed from all video recordings prior to transmission, and the analyzing technicians had no access to clinical or treatment-related information. Cadence symmetry index was defined as (affected steps per minute/unaffected steps per minute) × 100%; double support phase symmetry index was defined as the ratio of double support time on the affected side to that on the unaffected side, expressed in the same manner. Test–retest reliability was assessed in 10 pilot patients (ICC [intraclass correlation coefficient] > 0.80).[11]

Balance Function Parameters

The Berg balance scale (BBS) was employed to assess patients’ balance function.[12] All assessments were conducted by a trained rehabilitation therapist who was blinded to the treatment allocation of the patients. This scale comprises 14 items, each scored on a 0–4-point scale, yielding a total score of 0–56 points. Higher scores indicate better balance function. The BBS has demonstrated good test–retest reliability (ICC = 0.98) and concurrent validity with the postural assessment scale for stroke patients in the Chinese stroke population.[12] The lower-extremity subscale of the Fugl-Meyer Assessment (FMA-LE) was used to evaluate lower limb motor function. The maximum score on this subscale is 34 points, with higher scores reflecting better lower limb motor function.[13] The FMA-LE also shows excellent interrater reliability (ICC = 0.96) and internal consistency (Cronbach’s α = 0.89) in stroke patients.[13]

Walking Ability Parameters

The 10MWT was administered to measure the patient’s maximum walking speed. The test administrator was also blinded to group assignment. The 6MWT was used to assess walking endurance, with the maximum walking distance achieved within 6 min recorded in meters.[14]

Adverse Events

The occurrence of adverse events during the treatment period, including dizziness, lower-limb soreness or distension, and falls, was documented for both groups. The overall incidence of adverse events was subsequently calculated.

Statistical Analysis

Data were analyzed using SPSS 26.0 (IBM Corp., Armonk, NY, USA). Continuous variables are presented as mean ± SD, and categorical variables as n (%). Normality was assessed using the Shapiro–Wilk test (all P > 0.05), permitting parametric tests.

To compare treatment-related changes between groups, a linear mixed-effects model was used with group, time, and group × time interactions as fixed effects, adjusting for baseline outcome values. The group × time interaction was the primary test of the treatment effect. Post-treatment between-group comparisons were also performed using ANCOVA with baseline adjustment. Within-group changes were examined using paired t-tests, and post-treatment comparisons by independent t-tests.

For categorical variables, chi-square tests were used when expected frequencies were ≥5; otherwise, Fisher’s exact test was applied. The primary outcome (step symmetry index) was tested without multiplicity adjustment at α = 0.05 (two‑tailed). To control the false discovery rate for the secondary outcomes (gait symmetry: three indices; balance: two indices; walking: two indices), we applied the Benjamini–Hochberg procedure with a false discovery rate (FDR) of q = 0.05 as the primary adjustment method. For sensitivity analysis, we also computed the conventional Bonferroni‑corrected thresholds (domain‑wise α = 0.0167 for all secondary domains) to evaluate the robustness of the findings. Results from both the FDR and Bonferroni methods are reported in the respective tables. This dual‑reporting strategy ensures transparency regarding the influence of multiple comparison adjustments on statistical inferences.

To address potential confounding, multivariate linear regression was performed, adjusting for age, disease duration, Brunnstrom stage, and MMSE score. Propensity score matching was not feasible due to sample size limitations.

RESULTS

Comparison of Baseline Characteristics

No statistically significant differences were observed in the baseline demographic and clinical characteristics between the BWSTT-alone group and the RAS + BWSTT group (all P > 0.05). The details are presented in Table 1. Key prognostic variables including disease duration, Brunnstrom stage of the hemiplegic lower limb, and MMSE score were also compared between the two groups, and no significant differences were found (P > 0.05). To minimize potential selection bias inherent in the retrospective design, all eligible patients who met the inclusion criteria during the study period were consecutively enrolled.

Table 1.

Comparison of baseline characteristics between the two groups.

Index	BWSTT-alone Group (n = 55)	RAS + BWSTT group (n = 55)	t/χ2	P
Age (years)	58.72 ± 7.34	59.16 ± 6.89	0.324	0.747
Sex			1.784	0.162
Female	32 (58.18)	25 (45.45)
Male	23 (41.82)	30 (54.55)
Disease duration (months)	2.38 ± 0.85	2.57 ± 1.02	1.061	0.291
BMI (kg/m2)	24.15 ± 3.02	24.48 ± 3.11	0.565	0.573
Type of stroke			0.390	0.533
Ischemic stroke	37 (67.27)	40 (72.73)
Hemorrhagic stroke	18 (32.73)	15 (27.27)
Hemiplegic side			0.606	0.436
Left	35 (63.64)	31 (56.36)
Right	20 (36.36)	24 (43.64)
Brunnstrom stage (lower limb)			0.825	0.662
Stage III	18 (32.73)	15 (27.27)
Stage IV	22 (40.00)	24 (43.64)
Stage V	15 (27.27)	16 (29.09)
MMSE score	26.34 ± 1.52	26.58 ± 1.48	0.847	0.399
NIHSS score (baseline)	8.42 ± 2.35	8.67 ± 2.51	0.536	0.593
Lesion location			0.472	0.790
Cortical	20 (36.36)	18 (32.73)
Subcortical	28 (50.91)	30 (54.55)
Brainstem/cerebellar	7 (12.73)	7 (12.73)
Assistive device use at baseline			0.073	0.788
Yes	41 (74.55)	42 (76.36)
No	14 (25.45)	13 (23.64)

Note: Data for continuous variables are presented as mean ± SD; data for categorical variables are presented as n (%). Abbreviations: BMI: body mass index; MMSE: Mini-Mental State Examination. No statistically significant differences were observed between the two groups for any baseline variable (P > 0.05).

Comparison of Gait Symmetry Parameters

After 8 weeks of treatment, all three gait symmetry indices were significantly higher in both groups compared with baseline values. Furthermore, the three indices in the RAS + BWSTT group were significantly superior to those in the BWSTT-alone group (P < 0.05). Detailed data are presented in Table 2.

Table 2.

Comparison of gait symmetry parameters between the two groups before and after treatment.

Index	Time	BWSTT-alone group (n = 55)	RAS + BWSTT group (n = 55)	t	P
Step symmetry index	Pre-treatment	72.38 ± 8.26	71.55 ± 8.41	0.522	0.603
	After 8 weeks of treatment	76.26 ± 7.15a	78.74 ± 5.38a	2.055	0.042
Step frequency symmetry index	Pre-treatment	74.29 ± 7.83	72.86 ± 7.97	0.949	0.345
	After 8 weeks of treatment	77.17 ± 6.89a	80.08 ± 5.12a	2.514	0.013
Double support phase time proportion symmetry index	Pre-treatment	69.87 ± 9.12	70.52 ± 8.34	0.390	0.697
	After 8 weeks of treatment	74.35 ± 8.02a	78.63 ± 6.27a	3.118	0.002

Note: Compared with the same group before treatment ^aP < 0.05; For secondary outcomes, Benjamini–Hochberg FDR correction (q = 0.05) was applied as the primary adjustment; Bonferroni correction (α = 0.0167) was used for sensitivity. Under FDR, all three gait symmetry indices were significant (critical P-values: 0.0428 for i = 6 and 0.05 for i = 7). Under Bonferroni, the step symmetry index (P = 0.042) did not reach the corrected threshold. Group × time interaction P-values: step symmetry index P = 0.038; cadence symmetry index P = 0.021; double support phase time symmetry index P = 0.009.

To quantify the adjusted between‑group effects from the linear mixed‑effects model (adjusted for baseline values), the RAS + BWSTT group showed greater improvements than the BWSTT-alone group, with adjusted mean differences of 2.48 (95% confidence interval (CI): 0.38–4.58) for step symmetry index, 2.91 (95% CI: 0.65–5.17) for cadence symmetry index, and 4.28 (95% CI: 1.12–7.44) for double support phase time symmetry index. The group × time interaction P-values were 0.038, 0.021, and 0.009, respectively; after FDR correction, the corresponding P-values were 0.042, 0.032, and 0.018.

Comparison of Balance Function Parameters

After 8 weeks of treatment, both balance function scores were significantly higher in both groups compared with baseline values. Moreover, both scores in the RAS + BWSTT group were significantly superior to those in the BWSTT-alone group (P < 0.05). Detailed data are presented in Table 3.

Table 3.

Comparison of balance function parameters between the two groups before and after treatment (score).

Index	Time	BWSTT-alone group (n = 55)	RAS + BWSTT group (n = 55)	t	P
BBS rating	Pre-treatment	32.17 ± 5.82	30.93 ± 5.94	1.106	0.271
	After 8 weeks of treatment	35.26 ± 4.73a	37.52 ± 3.86a	2.745	0.007
FMA-LE score	Pre-treatment	19.82 ± 4.17	19.25 ± 4.23	0.712	0.478
	After 8 weeks of treatment	21.34 ± 3.58a	24.17 ± 3.02a	4.481	<0.001

Note: compared with the same group before treatment ^aP < 0.05; BBS: Berg balance scale; FMA-LE: Fugl–Meyer assessment lower extremity subscale. For secondary outcomes, Benjamini–Hochberg FDR correction (q = 0.05) was applied as the primary adjustment; Bonferroni correction (α = 0.0167) was used for sensitivity. Both BBS and FMA‑LE scores remained significant under both adjustment methods. Group × time interaction P-values: BBS score P = 0.013; FMA‑LE score P < 0.001.

For balance function, the adjusted mean differences for the RAS + BWSTT group relative to the BWSTT-alone group were 2.26 (95% CI: 0.63–3.89) for BBS score and 2.83 (95% CI: 1.48–4.18) for FMA‑LE score. The group × time interaction P-values were 0.013 and <0.001, respectively; after FDR correction, the corresponding P-values were 0.026 and <0.001.

Comparison of Walking Ability Parameters

After 8 weeks of treatment, both walking ability indices were significantly higher in both groups compared with baseline values. Additionally, both indices in the RAS + BWSTT group were significantly superior to those in the BWSTT-alone group (P < 0.05). Detailed data are presented in Table 4.

Table 4.

Comparison of walking ability parameters between the two groups before and after treatment.

Index	Time	BWSTT-alone group (n = 55)	RAS + BWSTT group (n = 55)	t	P
10MWT (m/s)	Pre-treatment	0.37 ± 0.12	0.36 ± 0.13	0.419	0.676
	After 8 weeks of treatment	0.45 ± 0.15a	0.52 ± 0.14a	2.530	0.013
6MWT (m)	Pre-treatment	182.53 ± 42.76	179.82 ± 43.15	0.331	0.741
	After 8 weeks of treatment	207.36 ± 38.29a	222.74 ± 35.61a	2.181	0.031

Note: Compared with the same group before treatment ^aP < 0.05; 10MWT: 10‑m walk test; 6MWT: 6‑min walk test. For secondary outcomes, Benjamini–Hochberg FDR correction (q = 0.05) was applied as the primary adjustment; Bonferroni correction (α = 0.0167) was used for sensitivity. Under FDR, the 6MWT result (P = 0.031, critical value for i = 6 = 0.0428) reached significance. Under Bonferroni, it does not reach the corrected threshold (required P < 0.0167) and is therefore interpreted as a non‑significant trend under that conservative method. Group × time interaction P-values: 10MWT P = 0.018; 6MWT P = 0.045.

For walking ability, the adjusted mean differences were 0.07 m/s (95% CI: 0.02–0.12) for 10MWT and 15.0 m (95% CI: 2.1–27.9) for 6MWT. The group × time interaction P values were 0.018 for 10MWT and 0.045 for 6MWT. Regarding the post‑treatment between‑group differences (unadjusted P = 0.013 for 10MWT and 0.031 for 6MWT), the Benjamini–Hochberg FDR procedure (q = 0.05) confirmed significance for both (critical value for 6MWT: 0.0428), whereas the Bonferroni correction (α = 0.0167) rendered the 6MWT nonsignificant (the 10MWT remained significant).

Comparison of Adverse Event Incidence

During the treatment period, the overall incidence of adverse events was 9.09% in the BWSTT-alone group and 5.45% in the RAS + BWSTT group. No statistically significant difference was observed in the incidence of adverse events between the two groups (P > 0.05). All reported adverse events were mild in severity, resolved spontaneously upon temporary suspension of training, and did not necessitate interruption of the subsequent treatment protocol. Detailed data are presented in Table 5.

Table 5.

Comparison of adverse event incidence between the two groups.

Group	n	Dizzy	Lower limb soreness	Fall down	Overall incidence
BWSTT-alone group	55	2 (3.64)	3 (5.45)	0 (0.00)	5 (9.09)
RAS + BWSTT group	55	1 (1.82)	2 (3.64)	0 (0.00)	3 (5.45)
Fisher/P					0.716

Note: Data are presented as n (%). BWSTT: body weight support treadmill training; RAS: rhythmic auditory stimulation. Fisher’s exact test was used for comparison between the two groups due to expected frequencies <5. P = 0.716.

DISSCUSSION

The present study revealed that, following treatment, the improvements in step length symmetry index, cadence symmetry index, and double support phase time symmetry index were significantly greater in the RAS + BWSTT group compared with the BWSTT-alone group. The observed improvements in gait symmetry may involve auditory–motor synchronization, but the present study provides no direct neurophysiological evidence. The proposed mechanisms (e.g., neural entrainment, central pattern generator modulation) are speculative and require confirmation in future studies using EEG, fMRI, or EMG. Concurrently, BWSTT reduces the weight-bearing load on the lower limbs and mitigates the fear of falling during ambulation, thereby providing a secure environment for repetitive gait practice.[15,16] The synergistic effect of the combined treatment may arise from bidirectional sensorimotor integration: rhythmic cues provide a predictable temporal template that reduces cortical processing demands for step timing, while body weight support allows the nervous system to focus on spatial aspects of gait symmetry without excessive compensatory efforts. In addition, the combined treatment allows patients to perform symmetrical stepping movements guided by rhythmic cues within a safe, weight-supported context. This approach facilitates the gradual correction of discrepancies in step length, cadence, and stance phase duration between the affected and unaffected limbs that are characteristic of hemiplegic gait. By fully leveraging the entrainment effects of RAS within the safe environment afforded by body weight support, this combined approach is positively correlated with multidimensional improvements in gait symmetry in patients with post-stroke hemiplegia. These findings provide a novel evidence base for optimizing clinical walking rehabilitation protocols following stroke.

The present study further indicated that patients in the RAS + BWSTT group achieved higher post-treatment BBS scores and FMA-LE scores than those in the BWSTT-alone group. From a neurophysiological perspective, RAS may standardize the motor sequencing of the trunk and lower limbs through constant rhythmic beat, possibly via the recruitment of the cerebello–thalamo–cortical pathway that governs the timing and coordination of postural adjustments. BWSTT facilitates the gradual execution of weight-shifting maneuvers and enhances the weight-bearing capacity and motor control of the affected lower limb.[17,18] The synchronous integration of these two therapies, which combines repetitive gait practice with rhythmic cueing, may contribute to strengthening lower limb muscle force and core stability, thereby improving both static and dynamic balance function. A previous study has indicated that RAS may improve dynamic balance control during walking in stroke patients, although its effect on static balance appears limited.[19] Another study suggested that while BWSTT effectively enhances lower limb motor function in stroke patients, improvements in balance function may exhibit a certain lag, with recovery of balance often occurring later than gains in lower limb muscle strength and walking capacity when this modality is applied alone.[20] By combining the temporal regulation provided by rhythmic cues with the muscle strengthening effects of weight-supported walking, the synchronous treatment approach employed in the present study achieved concurrent improvements in both static and dynamic balance. This strategy may shorten the recovery trajectory for balance function and offers a novel therapeutic perspective for the rehabilitation of balance deficits in patients with post-stroke hemiplegia.

In the present study, patients in the RAS + BWSTT group also demonstrated greater improvements in both the 10MWT and 6MWT compared with those in the BWSTT-alone group. Previous research has reported that improvements in gait symmetry may reduce energy expenditure during ambulation, whereas enhancements in balance function may augment both the safety and stability of walking.[21] RAS combined with BWSTT may simultaneously optimize gait symmetry, lower limb motor function, and balance function, thereby establishing a multidimensional foundation for the enhancement of walking capacity. Evidence indicates that conventional rehabilitation training supplemented with BWSTT may enhance walking speed and endurance in patients with post-stroke hemiplegia; however, its efficacy in ameliorating walking inefficiency attributable to gait abnormalities remains suboptimal.[22] It has been reported that RAS may improve the rhythmicity of gait but does not effectively address the limitations in walking capacity arising from insufficient lower-limb weight-bearing.[23,24] By adopting a synchronous treatment approach, the present study concurrently addressed three core deficits in walking rehabilitation among patients with post-stroke hemiplegia—namely, insufficient weight-bearing capacity, gait rhythm abnormalities, and poor gait symmetry—thereby achieving concurrent improvements in both walking speed and walking endurance. These findings provide an evidence-based rationale for the optimiz ation of clinical walking rehabilitation protocols.Furthermore, during the treatment period, no significant difference was observed in the overall incidence of adverse events between the two groups. No additional RAS-related adverse events were observed in this study. However, due to the limited total number of events (n = 8), no definitive safety conclusion can be drawn.

To further evaluate the clinical relevance of the findings, the observed between-group differences were compared with published minimal clinically important differences (MCIDs) for stroke populations.[25,26,27,28] The observed difference of 10MWT (0.07 m/s) and 6MWT (15 m) approached the lower limits of the reported MCID ranges (0.06–0.14 m/s and 14–34 m, respectively).[25,26] In contrast, the between-group differences in BBS (2.26 points) and FMA-LE (2.83 points) did not reach the corresponding MCID ranges (BBS 4–6 points and FMA-LE 4–5 points, respectively).[27,28] Therefore, although statistical significance was observed for some outcomes, the clinical meaningfulness of the between-group differences may be limited, particularly for balance and lower limb motor function. These findings should be interpreted with caution in clinical decision-making. For the 6MWT, although the between-group difference reached statistical significance under the Benjamini–Hochberg FDR correction (P = 0.031), it did not survive the more conservative Bonferroni correction (α = 0.0167). This result is sensitive to the choice of multiple comparison adjustment method. Furthermore, this improvement value falls at the lower bound of the published MCID range (14–34 m). Therefore, although this finding suggests a potential benefit in walking endurance, it should be interpreted with caution and requires confirmation in future prospective studies with larger sample sizes and pre-specified primary outcomes.

Limitations

Several limitations should be acknowledged. First, this retrospective, nonrandomized design is subject to inherent selection bias. Although multivariate adjustment was performed, residual confounding from unmeasured variables (e.g., baseline motivation, prior rehabilitation history, social support, prestroke hearing status, and baseline quantitative gait parameters) cannot be excluded. Second, the rehabilitation therapists delivering treatments were not blinded, potentially introducing performance bias; outcome assessors were blinded only during gait analysis. Third, the modest sample size, single-center setting, and 8-week follow-up limit generalizability and knowledge of long-term effects. Fourth, subgroup analyses by disease duration or Brunnstrom stage were not conducted, precluding identification of which patients benefitted most. Fifth, the mechanistic interpretations of RAS and BWSTT synergy are speculative, as no neurophysiological, imaging, or biomarker data were collected. Future prospective, multicenter randomized trials with larger samples, longer follow‑up, and a priori subgroup analyses are warranted.

CONCLUSION

RAS combined with BWSTT was associated with greater statistically significant improvements compared with BWSTT alone in improving gait symmetry, balance function, and walking ability in patients with post-stroke hemiplegia. However, for balance and lower limb motor function, the observed between-group improvements (2.26 points for BBS and 2.83 points for FMA-LE) did not reach the published MCID thresholds, suggesting that their clinical significance remains uncertain. This combined approach may provide a valuable reference for the development of clinical walking rehabilitation programs for patients with post-stroke hemiplegia. Additionally, the safety conclusion is limited by the small number of overall adverse events (n = 8) and the absence of systematic monitoring for hearing-related adverse effects (e.g., pure-tone audiometry or tinnitus assessment). Therefore, the safety profile of the combined treatment should be interpreted with caution.

Availability of Data and Materials

The datasets generated and/or analyzed during the current study are not publicly available due to institutional regulations and the protection of patient privacy and confidentiality, as the data contain potentially identifiable information extracted from hospital electronic medical records and rehabilitation assessment archives. However, de-identified participant data that support the findings of this study may be made available from the corresponding author upon reasonable request and subject to approval from the Ethics Committee of Changxing County Hospital of Traditional Chinese Medicine.

Author Contributions

Yan Xu: Conceptualization, Writing—original draft.

XiaoXia Ying: Funding acquisition, Resource allocation, Supervision, Writing—review & editing.

HaiJuan Lu: Data collection, Data organization.

DanDan Yu: Data collection, Data aggregation.

Fei Han: Data analysis and interpretation.

Ethics Approval and Consent to Participate

This study was approved by the Ethics Committee of Changxing County Hospital of Traditional Chinese Medicine (Ethics Approval No. 2026-018-01). The study protocol complied with the ethical principles outlined in the Declaration of Helsinki and relevant ethical and legal regulations. Due to the retrospective nature of this study, the requirement for informed consent was waived by the Ethics Committee of Changxing County Hospital of Traditional Chinese Medicine.

Conflicts of Interests

The authors declare that there are no conflicts of interest.

Funding Statement

None.