Motor function evaluation is a significant part of post-stroke rehabilitation protocols, and the evaluation of wrist motor function helps provide patients with individualized rehabilitation training programs. However, traditional assessment is coarsely graded, lacks quantitative analysis, and relies heavily on clinical experience. In order to objectively quantify wrist motor dysfunction in stroke patients, a novel quantitative evaluation system based on force feedback and machine learning algorithm was proposed. Sensors embedded in the force-feedback robot record the kinematic and movement data of the subject, and the rehabilitation doctor used an evaluation scale to score the wrist function of the subject. The quantitative evaluation models of wrist motion function based on random forest (RF), support vector machine regression (SVR), k-nearest neighbor (KNN), and back propagation neural network (BPNN) were established, respectively. To verify the effectiveness of the proposed quantitative evaluation system, 25 stroke patients and 10 healthy volunteers were recruited in this study. Experimental results show that the evaluation accuracy of the four models is all above 88%. The accuracy of BPNN model is 94.26%, and the Pearson correlation coefficient between model prediction and clinician scores is 0.964, indicating that the BPNN model can accurately evaluate the wrist motor function for stroke patients. In addition, there was a significant correlation between the prediction score of the quantitative assessment system and the physician scale score (p < 0.05). The proposed system enables quantitative and refined assessment of wrist motor function in stroke patients and has the feasibility of helping rehabilitation physicians in evaluating patients’ motor function clinically.

During upper limb rehabilitation training and assessment, compensatory movements can significantly impede the recovery of motor function in stroke survivors. Current methods for detecting these movements primarily rely on visual sensors, torso restraint belts, and pressure pads. However, these approaches present challenges such as obstacle occlusion, complex operation, and an inability to provide a quantitative assessment of compensatory movements. To address these issues, we developed a system for detecting and quantitatively evaluating compensatory movements, utilizing wearable sensors and machine learning algorithms. We recruited 30 healthy volunteers and 32 stroke patients to complete seven experimental motions. Their movement data was recorded using three inertial measurement units placed on the upper limbs and sternum. After preprocessing the data, six machine learning algorithms (KNN, SVM, MLP, RF, XGBoost, and LightGBM) were employed to train a multi-label classification model for detecting and quantitatively evaluating compensatory movements. The algorithm with the highest performance was selected as the system’s classifier. The results show that the proposed system improves the number and accuracy of compensatory movement detection and achieves accurate classification and detection of 7 compensatory movements (average accuracy is 95.7%, F1-score is 96.8%). Furthermore, KNN classifier classified compensatory movements as mild, moderate, and severe according to the different severity of the compensatory movements and obtained an average accuracy of 93.7%, which proved the reliability and responsiveness of the system. This system is applicable to stroke patients in clinical, home, and community-based rehabilitation settings, enabling a quantitative assessment of compensatory movements.

Background/Objectives: With the advancement of intelligent sensing technology, rehabilitation systems based on wearable devices have a positive impact on the functional recovery and quality of life of stroke patients. This study aims to evaluate the application value of a contextualized intelligent rehabilitation training system for stroke survivors, which is based on wearable devices, in the rehabilitation of motor function impairments following stroke. Methods: A randomized controlled trial design was employed, in which 100 stroke patients were randomly divided into a control group (n = 50, receiving standard physical therapy rehabilitation interventions) and an experimental group (n = 50). The experimental group additionally underwent motor function rehabilitation interventions and intelligent assessments through a wearable device-based contextual intelligent rehabilitation training system, with sessions conducted twice daily for 30 min each, five days a week, over a duration of eight weeks. Both groups of patients underwent clinical scale evaluations and data collection before and after the treatment, with primary outcome measures including motor ability (Fugl–Meyer Assessment, FMA), activities of daily living (Modified Barthel Index, MBI), and participation in rehabilitation therapy. The intervention effects of both groups were compared after eight weeks of rehabilitation. Results: Prior to the intervention, there were no significant differences in Fugl–Meyer Assessment (FMA) and Modified Barthel Index (MBI) scores between the experimental group and the control group (p > 0.05). After eight weeks of rehabilitation, the experimental group demonstrated significantly superior performance in motor function (FMA scores) and activities of daily living (MBI scores) compared to the control group (p < 0.01). Conclusions: The intelligent rehabilitation system significantly enhances motor function and activities of daily living in stroke survivors. Compared to traditional rehabilitation methods, it improves patient adherence to rehabilitation training and overall outcomes.