Perceptual Feedback through Multisensory Fusion in Hand Function Rehabilitation by A Machine Learning Approach

Patients with impaired hand function often experience both motor and sensory deficits. Previous research on hand function rehabilitation has predominantly focused on passive rehabilitation training with external devices, neglecting the necessity of establishing channels for regulation and sensory feedback during the rehabilitation process. These studies seldom utilize feedback information to present real-time status of rehabilitation training to users, overlooking a crucial factor of active patient participation in rehabilitation. In this study, we employed a machine learning model (Long Short-Term Memory) to fuse hand movement status and grip strength information collected from multiple sensors, thereby enhancing the perceptual feedback system, and improve the perceptual abilities of patients with hand dysfunction during the use of rehabilitation exoskeleton devices, ultimately enhancing the efficiency of neurological rehabilitation.