Innovative Applications of Multimodal Sensing Technology in Sports Rehabilitation Assessment and Training

Abstract

This study explores an innovative approach to evaluating the training effectiveness of lower limb exoskeleton robots by integrating multiple data types, including electrophysiological signals and kinematic measurements, to assess patients' walking ability quantitatively. Through precisely defined synergistic indicators, this method effectively combines different types of data and dramatically improves the efficiency and accuracy of rehabilitation assessment. First, the patient's lower extremity electro myoelectric activity and movement data were recorded while walking with exoskeleton assistance. Secondly, the key EMG and kinematic features are analyzed and extracted by a collaborative quantization algorithm based on the theory of muscle cooperative work. Then, this information from different levels is integrated to build a feature fusion model, based on which the lower limb motor function score is calculated. The development of multi-channel lower limb exoskeleton human-computer interaction technology for sports training can provide a variety of standardized and standardized auxiliary training for athletes and meet the human body's multi-sensory immersion. A multi-step, multi-degree-of-freedom motion planning algorithm is proposed to reproduce various activities the human body requires. Secondly, the lower extremity-oriented multi-modal human-computer interaction technology is studied to realize the display and guidance of standard movement in information space on the virtual reality competition training simulation platform. Build a motion database to assist and correct basic motion in physical space. The experimental results showed a significant correlation between the extracted myoelectric and kinematic synergistic features and the clinical evaluation tools, with the correlation coefficients reaching 0.832 and 0.859, respectively. The fusion features show a stronger correlation when applying the K-nearest neighbor (KNN) algorithm. This evaluation method cannot only optimize the training strategy of the exoskeleton robot according to the results but also provide the possibility to realize the "man in the ring" mode of evaluation and training simultaneously.