Health-convergence-based muscle synergy variability for quantifying neuromuscular control responses evoked by upper limb rehabilitation robot training in stroke patients

 With NO protocol created and distributed to all 10 million yearly survivors; COMPLETELY FUCKING USELESS!

Health-convergence-based muscle synergy variability for quantifying neuromuscular control responses evoked by upper limb rehabilitation robot training in stroke patients

• Ye Zhou,
• Xiaoying Wu,
• Haimei Zhou,
• CengCeng Xie,
• Yu Tang,
• Tian Shu,
• Xin Zhang,
• Xing Wang,
• Wanling Jiang &
• Wensheng Hou 

We’re sharing this article early to provide faster access to peer-reviewed, accepted research. It is citable and carries a permanent DOI. This version is subject to further edits and will be replaced automatically by the final Version of Record. All legal disclaimers apply.

Abstract

Background

Upper limb rehabilitation robot-assisted training (UE-RAT) facilitates post-stroke motor recovery. Its efficacy is evaluated using clinical scales. However, these scales mainly reflect overall motor impairment and functional performance, lacking indices to assess patients’ neuromuscular control during training, thereby limiting precise evaluation of training effects. Therefore, this study quantified neuromuscular control in stroke patients through muscle synergy analysis with fine-grained movement decomposition and proposed a muscle synergy variability index based on healthy synergy templates to measure convergence toward healthy neuromuscular control.

Methods

Eighteen age-matched healthy subjects and 62 stroke patients were recruited. Patients underwent 4-week UE-RAT in assistive, active, or resistive modes according to their functional impairment. Electromyographic and kinematic data were synchronously recorded from the affected upper limb weekly. Continuous tasks were segmented into four specific movement types based on motion trajectories. Muscle synergy patterns were extracted from healthy subjects for each movement segment and clustered to construct reference synergies. Using this benchmark, the similarity and coverage of patients’ single-movement synergies relative to reference synergies were calculated to derive the variability index (V_variability). Finally, standardized by healthy controls, the neuromuscular control deviation score (DS) was quantified for each training mode.

Results

The proposed V_variability significantly distinguished patients from healthy controls across all training modes (P < 0.05). Longitudinal observation demonstrated average reductions in V_variability of 5.7%-11.2%, 13.2%-25.5%, and 14.0%-25.1% after training in the assistive, active, and resistive groups, respectively (all P < 0.05). After adjustment for clinical covariates, DS remained significantly different across training modes (P < 0.05), with the highest deviations in the severely impaired assistive group (0.79–1.19) and the lowest in the least impaired resistive group (0.26–0.42). DS was significantly negatively correlated with FMA-UE scores (r= -0.317 to -0.526) and exhibited stronger correlations than conventional synergy-based metrics, including synergy structural similarity (r = 0.020–0.448) and synergy merging (r= -0.171 to -0.408), supporting its potential as a complementary quantitative measure of neuromuscular control deviation during robotic rehabilitation.

Conclusion

The proposed V_variability can quantify deviations of stroke patients’ neuromuscular responses from healthy patterns during UE-RAT, providing an objective tool to assess task suitability and supporting neuromuscular control-guided precision rehabilitation.