Effective unilateral/bilateral robot-assisted training for upper limb motor function rehabilitation: a cross-sectional study

Guang Feng, Guohong Chai, Jiaji Zhang, Tao Song, Changcheng Shi, Jialin Xu, Guokun ZuoFrontiers in Human Neuroscience 2025 · DOI

Abstract

The current outcomes provide valuable guidelines for designing individualized robot-assisted training protocols, potentially promoting the clinical rehabilitation effect.

Introduction

Stroke remains a primary cause of upper limb motor impairment globally. While robot-assisted rehabilitation provides high-intensity, objective therapy, its clinical superiority over traditional methods remains a subject of debate. This lack of consensus often stems from a deficiency in standardized, patient-tailored protocols. Specifically, the interactive mechanisms between training modes (unilateral vs. bilateral), task complexity, and feedback modalities (visual vs. force) are not fully understood. This study quantifies these variables to establish a foundation for individualized rehabilitation strategies.

Methodology

The study utilized a custom-developed Bilateral Upper Limb Rehabilitation Robot (BULRR). To establish a performance baseline, 18 healthy, right-handed subjects (mean age 27) participated in trials across three distinct training modes:

  1. Unilateral Passive Training (UPT): The robot guides the "impaired" limb along a predefined path.
  2. Bilateral Passive Training (BPT): The subject uses their healthy limb to drive the robot, which then mirrors the movement to the "impaired" limb.
  3. Unilateral Active Training (UAT): The subject moves the limb actively against virtual damping.
The schematic diagram of the BULRR and the experimental setup for robot-assisted rehabilitation training and evaluation.

Participants performed three virtual tasks—straight-line, circular, and S-shaped—under four feedback conditions: Visual, Force, Visual-Force (multi-modality), and None. Researchers monitored Tracking Error (TE), Interactive Force (IF), and muscle activation via surface electromyography (sEMG) across five major arm muscles. Mental burden was assessed using the NASA-TLX scale.

Key Insights and Results

The Bilateral Advantage

Bilateral passive training (BPT) combined with visual feedback elicited significantly higher muscle activation (0.63 ± 0.26) compared to unilateral passive training (0.24 ± 0.05). This supports the theory that bilateral movement helps "prime" the damaged hemisphere of the brain through interhemispheric pathways, potentially enhancing neuroplasticity.

The "Feedback Paradox"

While multi-modality feedback (visual + force) resulted in the lowest tracking errors (6.6 mm in circular tasks) and the lowest mental burden, it also led to lower muscle activation. In contrast, UAT with single-modality feedback (visual or force) encouraged higher active participation (0.78 ± 0.12 vs. 0.53 ± 0.13). This suggests that excessive guidance may make tasks "too easy," inadvertently reducing the active effort necessary for recovery.

Complexity and Engagement

Simple straight-line tasks resulted in low engagement. Circular and S-shaped tasks significantly enhanced the benefits of the training strategies, indicating that path unpredictability is a crucial element for effective rehabilitation. Furthermore, while visual feedback helped subjects adjust to trajectories more easily, force feedback was superior at preventing large deviations from the path.

Conclusion and Future Directions

This study provides a benchmark for coordination in healthy individuals, but its application to clinical populations requires further investigation. Stroke patients exhibit different neurophysiology and cognitive loads, which may alter their response to multi-modality feedback.

Future research should:

  • Evaluate these protocols in stroke patients with varying degrees of impairment.
  • Incorporate real-time cortical activation monitoring via EEG, fNIRS, or fMRI.
  • Directly compare robot-assisted outcomes against the Fugl-Meyer Assessment (FMA-UE) clinical standard.

Summary updated

Abstract

The current outcomes provide valuable guidelines for designing individualized robot-assisted training protocols, potentially promoting the clinical rehabilitation effect.
BibTeX
@inproceedings{feng2025effective,
  author = {Guang Feng and Guohong Chai and Jiaji Zhang and Tao Song and Changcheng Shi and Jialin Xu and Guokun Zuo},
  title = {{Effective unilateral/bilateral robot-assisted training for upper limb motor function rehabilitation: a cross-sectional study}},
  booktitle = {Frontiers in Human Neuroscience},
  year = {2025},
  doi = {10.3389/fnhum.2025.1571624},
}
Figure