EMG-based Assessment Device for Hand Rehabilitation with Cloud Analysis

Stroke has become a prevalent cardiovascular ailment that impacts human lives due to aging, chronic health issues, or injuries. Often, stroke survivors require rehabilitation to regain muscle function and relearn their motor skills. Unfortunately, the insufficient availability of rehabilitation services and a shortage of practitioners impede their recovery progress. Therefore, the development of an assessment device for hand rehabilitation based on electromyography (EMG) is proposed. The main challenge is to design an assessment device that can recognize the user’s motion intention, which can be done by utilizing the electromyogram signals generated by forearm muscles contributed by the movement and/or grasping abilities of the hand. In this research, an EMG-mechanical sensor fusion is designed which combines an electronic conditioning circuit that measures the EMG signals extracted from the forearm muscles with mechanical sensor modalities (self-assembled hand dynamometer and flex sensor for wrist angle measurement). Once the proof of concept is established, the designed system is interfaced with the data analytics platform. This platform stores the collected data in the cloud, making it accessible for rehabilitation assessment. Scikit-Fuzzy with multiple linear regression is used to model the relationship between EMG signals, handgrip force, and wrist angle and map it to the hand assessment score chart. Based on the analysis, the trained model using Fuzzy with multiple linear regression achieved a 73.32% prediction accuracy. The proposed research contributes towards the aim of providing better rehabilitation diagnostic for post-stroke hand impairment survivors in regaining their hand strength and functionality and improving their quality of life.