Affiliations: [a] Institute of Rehabilitation Engineering and Technology, School of Health Science and Engineering, University of Shanghai for Science and Technology, Shanghai, China | [b] Shanghai Engineering Research Center of Assistive Devices, University of Shanghai for Science and Technology, Shanghai, China | [c] Changhai Hospital, Shanghai, China | [d] Third Affiliated Hospital of Naval Medical University, Shanghai, China
Correspondence:
[*]
Corresponding authors: Sujiao Li and Hongliu Yu, Institute of Rehabilitation Engineering and Technology, University of Shanghai for Science and Technology, No. 516 Jungong Rd, Shanghai, China. E-mail: sujiao.li@usst.edu.cn and yhl_usst@outlook.com.
Abstract: BACKGROUND: Rehabilitation assessment is a critical component of rehabilitation treatment. OBJECTIVE: This study focuses on a comprehensive analysis of patients’ movement performance using the upper limb rehabilitation robot. It quantitatively assessed patients’ motor control ability and constructed an intelligent grading model of functional impairments. These findings contribute to a deeper understanding of patients’ motor ability and provide valuable insights for personalized rehabilitation interventions. METHODS: Patients at different Brunnstrom stages underwent rehabilitation training using the upper limb rehabilitation robot, and data on the distal movement positions of the patients’ upper limbs were collected. A total of 22 assessment metrics related to movement efficiency, smoothness, and accuracy were extracted. The performance of these assessment metrics was measured using the Mann-Whitney U test and Pearson correlation analysis. Due to the issue of imbalanced sample categories, data augmentation was performed using the Synthetic Minority Over-sampling Technique (SMOTE) algorithm based on weighted sampling, and an intelligent grading model of functional impairment based on the Extreme Gradient Boosting Tree (XGBoost) algorithm was constructed. RESULTS: Sixteen assessment metrics were screened. These metrics were effectively normalized to their maximum values, enabling the derivation of quantitative assessment scores for motor control ability across the three dimensions through a weighted fusion approach. Notably, when applied to the data-enhanced dataset, the intelligent grading model exhibited remarkable improvement, achieving an accuracy rate exceeding 0.98. Moreover, significant enhancements were observed in terms of precision, recall, and f1-score. CONCLUSION: The research findings demonstrate that this study enables the quantitative assessment of patients’ motor control ability and intelligent grading of functional impairments, thereby contributing to the efficiency enhancement of clinical rehabilitation assessment. Moreover, this method resolves the issues associated with the subjectivity and prolonged periods of traditional rehabilitation assessment methods.
