Preliminary effects of robotic-assisted gait training in post-stroke patients: A pilot study

Authors

  • Iyyappan ManickavasagamDepartment of Neurosciences, Saveetha College of Physiotherapy, Saveetha Institute of Medical and Technical Sciences, Chennai, Tamil Nadu, India
  • Vignesh SrinivasanDepartment of Neurosciences, Saveetha College of Physiotherapy, Saveetha Institute of Medical and Technical Sciences, Chennai, Tamil Nadu, India
  • Yamini UmasankarDepartment of Neurosciences, Saveetha College of Physiotherapy, Saveetha Institute of Medical and Technical Sciences, Chennai, Tamil Nadu, India
  • Jagatheesan AlagesanDepartment of Paediatrics, School of Paramedical Allied and Health Care Sciences, Mohan Babu University, Tirupati, India

DOI:

Keywords

assistive technology, disability, health, quality of life, stroke rehabilitation

Downloads

Correspondence

Vignesh Srinivasani
Email: vigneshphysio1989@gmail.com

Publication history

Received: 24 Sep 2025
Accepted: 25 Dec 2025
Published online: 29 Dec 2025

Funding

None

Ethical approval

Approved by Dubai scientific research ethics committee (Ref. no.: DSREC-SR-10/2024_02, Dated 11 Nov 2024)

Trial registration number

Not available

Copyright

© The Author(s) 2025; all rights reserved. 
Published by Bangladesh Medical University (former Bangabandhu Sheikh Mujib Medical University)
Key messages
Restoration of gait is central to post-stroke independence and functional recovery. This small pilot study demonstrates that robotic-assisted gait training adds value to maintaining balance, motor function and mobility compared with conventional therapy in the rehabilitation of people with stroke. However, adequately powered studies for longer-term follow-up are warranted.
Stroke remains a leading cause of long-term disability worldwide, with gait impairment and balance dysfunction being among the most disabling sequelae affecting independence and quality of life [1, 2]. Restoration of walking ability is therefore a primary goal of post-stroke rehabilitation. Conventional physiotherapy remains the cornerstone of gait rehabilitation; however, variability in training intensity, therapist fatigue, and limited opportunities for high-repetition, task-specific practice may limit optimal recovery [3]. Robotic-assisted gait training (RAGT) has emerged as a promising adjunct, enabling repetitive, controlled and task-oriented walking practice in a safe and standardised environment [4, 5]. Despite growing interest, evidence regarding its feasibility and preliminary clinical effects in routine rehabilitation settings remains heterogeneous, particularly in low- and middle-income contexts. This research letter reports the preliminary findings of a randomised pilot study comparing robotic-assisted gait training with conventional physiotherapy in individuals recovering from stroke.

This hospital-based pilot randomised study was conducted at a neurorehabilitation centre, with institutional ethical approval obtained in line with the declaration of Helsinki. Written informed consent was obtained from all participants before enrollment. Ten individuals with first-ever ischemic or hemorrhagic stroke, aged 30–60 years, were recruited and randomly allocated to either robotic-assisted gait training (RAGT group, n=5) or conventional physio-therapy (control group, n=5) using a sealed-envelope method. All participants had mild to moderate lower-limb spasticity (Modified Ashworth Scale ≤ 2), Brannstrom stages II–IV, Mini-Mental State Examination scores ≥ 24 and functional ambulation category scores between 2 and 4. No participants were lost to follow-up.

The RAGT group received treadmill-based robotic gait training using the Lokomat system with partial body-weight support, three sessions per week for 12 weeks. Training parameters such as walking speed and body-weight support were progressively adjusted according to participant tolerance and clinical judgment. Each session lasted approximately 40 minutes and focused on repetitive, task-specific gait practice under guided robotic assistance. The control group received conventional physiotherapy three times per week, consisting of breathing exercises, facilitatory techniques, bed mobility training, lower-limb stretching, balance activities and overground gait training, with progressive advancement over the intervention period. All interventions were delivered by licensed physiotherapists experienced in neurorehabilitation. No adverse events were reported in either group.

Balance, motor recovery and gait performance were assessed at baseline and after 12 weeks using validated clinical outcome measures: The Berg Balance Scale, Fugl–Meyer Assessment for Lower Extremity, Functional Gait Assessment and Timed Up and Go test. Data were analysed using Wilcoxon signed-rank test for within-group changes (P <0.01) and Mann–Whitney U test for between-group differences (P < 0.05).


Both groups demonstrated statistically significant improvements across all outcome measures over the 12-week intervention period. However, participants in the RAGT group consistently showed greater preliminary mean improvements in balance, motor recovery and functional mobility compared with those receiving conventional physiotherapy. The most notable between-group differences favored RAGT for Berg Balance Scale, lower-limb motor recovery Fugl–Meyer Assessment for Lower Extremity, dynamic gait performance and functional mobility Timed Up and Go test. A consolidated summary of outcome changes is presented in Table 1.

Categories

Number (%)

Sex

 

   Male

36 (60.0)

   Female

24 (40.0)

Age in yearsa

8.8 (4.2)

   Education

 

   Pre-school

20 (33.3)

   Elementary school

24 (40.0)

   Junior high school

16 (26.7)

Cancer diagnoses

 

   Acute lymphoblastic leukemia

33 (55)

   Retinoblastoma

5 (8.3)

   Acute myeloid leukemia

4 (6.7)

   Non-Hodgkins lymphoma

4 (6.7)

   Osteosarcoma

3 (5)

   Hepatoblastoma

2 (3.3)

   Lymphoma

2 (3.3)

   Neuroblastoma

2 (3.3)

   Medulloblastoma

1 (1.7)

   Neurofibroma

1 (1.7)

   Ovarian tumour

1 (1.7)

   Pancreatic cancer

1 (1.7)

   Rhabdomyosarcoma

1 (1.7)

aMean (standard deviation)

Categories

Number (%)

Sex

 

   Male

36 (60.0)

   Female

24 (40.0)

Age in yearsa

8.8 (4.2)

Education

 

   Pre-school

20 (33.3)

   Elementary school

24 (40.0)

   Junior high school

16 (26.7)

Cancer diagnoses

 

Acute lymphoblastic leukemia

33 (55)

Retinoblastoma

5 (8.3)

Acute myeloid leukemia

4 (6.7)

Non-Hodgkins lymphoma

4 (6.7)

Osteosarcoma

3 (5)

Hepatoblastoma

2 (3.3)

Lymphoma

2 (3.3)

Neuroblastoma

2 (3.3)

Medulloblastoma

1 (1.7)

Neurofibroma

1 (1.7)

Ovarian tumour

1 (1.7)

Pancreatic cancer

1 (1.7)

Rhabdomyosarcoma

1 (1.7)

aMean (standard deviation)

Table 1 Comparison of mean changes (post minus pre) between robot-assisted gait training and conventional therapy according to outcome measures

Topic

Teaching method

Pre-test score

Post-test score

Difference (post minus pre-test scores)

P

 

Cerebellar disorder 

Traditional teaching

1.2 (0.9)

2.8 (1.0)

1.6

<0.001

ECE

1.2 (0.9)

6.4 (1.4)

5.2

<0.001

Parkinson's disease 

Traditional teaching

1.2 (0.9)

1.9 (0.7)

0.7

<0.001

ECE

1.2 (0.9)

6.0 (1.3)

4.8

<0.001

ECE indicates early clinical exposure. All values expressed as mean (standard deviation)

Topic

Teaching method

Pre-test score

Post-test score

Difference between post and pre-test score

P

(Post vs pre)

Cerebellar Disorder

 

Traditional Teaching

1.2 (0.9)

2.8 (1.0)

-1.6

<0.001

ECE

1.2 (0.9)

6.4 (1.4)

-5.2

<0.001

Parkinson's Disease

 

Traditional Teaching

1.2 (0.9)

1.9 (0.7)

-0.7

<0.001

ECE

1.2 (0.9)

6.0 (1.3)

-4.8

<0.001

ECE indicates Early clinical exposure. All values expressed as mean (standard deviation)

Variables

Results

Number (%)

Sex

Male

29 (48.0)

Female

31 (52.0)

Symptoms 

 

Dyspnea

56 (93.3)

Chest pain

55 (91.7)

Fatigue

55 (91.7)

Leg oedema

47 (78.3)

Palpitation

49 (81.6)

Cough

27 (45.0)

Hoarseness of voice

3 (5.0)

Hemoptysis

5 (8.3)

Syncope

3 (5.0)

Cyanosis

3 (5.0)

Severity of pulmonary hypertension  

Mild

28 (46.7)

Moderate

22 (36.7)

Severe

10 (16.7)

Types of pulmonary hypertension

Type 1

19 (31.6)

Type 2

27 (45.0)

Type 3

7 (11.7)

Type 4

3 (5.0)

Type 5

4 (6.7)

Mean (SD)

Age (years)

49.6 (14.6)

RVSP (mmHg) by severity of pulmonary hypertension    

Mild

43 (7.0)

Moderate

58 (5.0)

Severe

80 (7:0)

mPAP (mmHg) of type of pulmonary hypertension   

Type 1

54 (28.9)

Type 2

46 (7.6)

Type 3

47.4 (11.0)

Type 4

53 (2.0)

Type 5

49.5 (16.2)

SD indicates standard deviation; RVSP, right ventricular systolic pressure; mPAP, mean pulmonary artery pressure. RVSP 36–49 mmHg is mild, 50 –69 mmHg is moderate, and ≥70 mmHg is severe pulmonary hypertension. Types of pulmonary hypertension are based on etiology as per reference.

Category

Key Factors

Weight

Strengths

Strong management support, skilled workforce, compliance with legal regulations

0.338

Weaknesses

Logistical complexity, inadequate segregation, financial constraints

0.13

Opportunities

Industry collaboration, environmental policies, new technology

0.094

Threats

Limited space, lack of coordination, high investment risk

0.329

Pain level

Number (%)

P

Pre

Post 1

Post 2

Mean (SD)a pain score

4.7 (1.9)

2.7 (1.6)

0.8 (1.1)

<0.001

Pain categories

    

   No pain (0)

-

1 (1.7)

31 (51.7)

<0.001

   Mild pain (1-3)

15 (25.0)

43 (70.0)

27 (45.0)

 

   Moderete pain (4-6)

37 (61.7)

15 (25.0)

2 (3.3)

 

   Severe pain (7-10)

8 (13.3)

2 (3.3)

-

 

aPain scores according to the visual analogue scale ranging from 0 to 10; SD indicates standard deviation

Outcome measure

Group for each (n=5)

Mean change (95% confidence interval) a

Berg balance scale

Robot-assisted gait training

21.2 (18.7–23.7)

Conventional

5.4 (4.0–6.8)

Fugl–Meyer assessment

Robot-assisted gait training

6.6 (6.1–7.3)

Conventional

2.4 (1.9–2.9)

Functional gait assessment

Robot-assisted gait training

4.4 (4.1–4.7)

Conventional

1.6 (1.4–1.8)

Timed up and go

Robot-assisted gait training

8.1 (7.5–8.7)

Conventional

3.7 (3.4–4.0)

a All P values for within group comparisons are significant at 1% level (according to the Wilcoxon signed-rank test). Between-group mean changes are significantly different at the 5% level (Mann–Whitney U test).

These preliminary findings suggest that robotic-assisted gait training may offer additional benefits over conventional physiotherapy in improving balance, gait and mobility following stroke. The observed improvements may be attributed to the high-repetition, task-specific nature of robotic training, which facilitates motor relearning and may enhance neuroplastic adaptation through consistent sensory feedback and guided movement patterns [6, 7]. Previous studies have highlighted the importance of intensive, repetitive gait practice in promoting functional walking recovery after stroke, and robotic systems provide a practical means of delivering such training with reduced therapist burden [8].

Improvements in balance, as reflected by higher Berg Balance Scale scores in the RAGT group, are clinically meaningful, given the strong association between balance deficits and fall risk in stroke survivors [9]. Similarly, greater gains in Fugl–Meyer Assessment for Lower Extremity scores indicate enhanced lower-limb motor recovery, likely driven by repetitive practice of near-normal gait kinematics. Enhanced Functional Gait Assessment and reduced Timed Up and Go test times further suggest superior improvements in dynamic gait control and functional mobility, which are essential for safe community ambulation.


Robotic-assisted gait training improved balance, motor recovery, and gait in individuals with stroke more effectively than conventional physiotherapy. Nevertheless, these findings must be interpreted cautiously. Given the pilot study's small sample size, the results are exploratory and not powered to establish definitive efficacy. The absence of a blinding and reliance on clinical outcome measures limit generalisability. Additionally, long-term retention of gains and cost-effectiveness of robotic-assisted training were not evaluated. Despite these limitations, the study demonstrates feasibility and safety, thereby justifying adequately powered randomised controlled trials.

Acknowledgements
We sincerely acknowledge the developers of clinical measures for providing validated tools that made this research possible. We extend our heartfelt gratitude to all the patients who participated in this study for their time, cooperation, and commitment. We also thank our co-authors and research team members for their valuable guidance, support, and contributions throughout the study.
Author contributions
Conception or design of the work; or the acquisition, analysis, or interpretation of data for the work: IM, YU. Drafting the work or reviewing it critically for important intellectual content: IM, VS, YU, JA. Final approval of the version to be published: IM, VS, YU, JA. Accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved: IM, VS, YU, JA.
Conflict of interest
We do not have any conflict of interest.
Data availability statement
We confirm that the data supporting the findings of the study will be shared upon reasonable request.
Supplementary file
None
    References
    1. Murphy SJ, Werring DJ. Stroke: Causes and clinical features. Medicine: Acute neurology. 2020 Sep 1;48(9):561-566. https://doi.org/10.1016/j.mpmed.2020.06.002
    2. Jambi LK, Hamad A, Salah H, Sulieman A. Stroke and disability: Incidence, risk factors, management and impact. Journal of Disability Research. 2024 Sep 5;3(7):20240094. doi: https://doi.org/10.57197/JDR-2024-0094
    3. Latham NK, Jette DU, Slavin M, Richards LG, Procino A, Smout RJ, Horn SD. Physical therapy during stroke rehabilitation for people with different walking abilities. Archives of physical medicine and rehabilitation. 2005 Dec 1;86(12):41-50. https://doi.org/10.1016/j.apmr.2005.08.128
    4. Bonnyaud C, Zory R, Boudarham J, Pradon D, Bensmail D, Roche N. Effect of a robotic restraint gait training versus robotic conventional gait training on gait parameters in stroke patients. Experimental Brain Research. 2014 Jan;232(1):31-42. doi: https://doi.org/10.1007/s00221-013-3717-8
    5. Zhu YH, Ruan M, Yun RS, Zhong YX, Zhang YX, Wang YJ, Sun YL, Cui JW. Is Leg-Driven Treadmill-Based Exoskeleton Robot Training Beneficial to Poststroke Patients: A Systematic Review and Meta-analysis. American Journal of Physical Medicine & Rehabilitation. 2023 Apr 1;102(4):331-339. doi: https://doi.org/10.1097/PHM.0000000000002098
    6. Jones TA. Motor compensation and its effects on neural reorganisation after stroke. Nature Reviews Neuroscience. 2017 Mar 23;18(5):267-280. doi: https://doi.org/10.1038/nrn.2017.26
    7. Pichardo A, Malovec M. Enhancing Post-Stroke Gait Rehabilitation with Robot-Assisted Therapy: A Focus on Step Repetitions and Neuroplasticity. Archives of Clinical and Biomedical Research. 2025;9:212-219. doi: https://doi.org/10.26502/acbr.50170453
    8. Nadeau SE, Wu SS, Dobkin BH, Azen SP, Rose DK, Tilson JK, Cen SY, Duncan PW; LEAPS Investigative Team. Effects of task-specific and impairment-based training compared with usual care on functional walking ability after inpatient stroke rehabilitation: LEAPS Trial. Neurorehabil Neural Repair. 2013 May;27(4):370-380. doi: https://doi.org/10.1177/1545968313481284
    9. Chen K, Zhu S, Tang Y, Lan F, Liu Z. Advances in balance training to prevent falls in stroke patients: A scoping review. Front Neurol. 2024 Feb 5;15:1167954. doi: https://doi.org/10.3389/fneur.2024.1167954