Design of Intelligent Rehabilitation Evaluation Scale for Stroke Patients Based on Genetic Algorithm and Extreme Learning Machine

Abstract

The rehabilitation of stroke patients is a long-term process. To realize the automation and quantification of upper limb rehabilitation assessment of stroke patients, an automatic prediction model of rehabilitation evaluation scale was established by extreme learning machine (ELM) according to Fugl-Meyer motor function assessment (FMA). Four movements in the shoulder and elbow joints of FMA were selected. Two acceleration sensors fixed on the forearm and upper arm of the hemiplegic side were used to collect the motion data of 35 patients. After preprocessing and feature extraction, the feature selection was carried out based on genetic algorithm and ELM, and the single-action model and comprehensive prediction model were established, respectively. The results show that the model can accurately and automatically predict the shoulder and elbow score of FMA, and the root mean square error of prediction is 2.16. This method breaks through the limitations of subjectivity, time-consuming and dependence on rehabilitation doctors in the traditional evaluation. It can be easily used in the assessment of long-term rehabilitation.