AI Classification System on Sarcopenia for Elderly

Abstract

The world has gradually entered an aging society, and many older people die of falls every year with sarcopenia being one of the main reasons for the elderly to fall. Thus, we present a novel approach with an intelligent rehabilitation knee brace developed by a Taiwanese start-up company (Ai Free) which collected 755 data from 55–70 age older patients in a local Tainan community in Taiwan. EMG signals and six-axis sensor values were extracted from the patients. According to the root mean square (RMS) value for muscle strength, the mean frequency (MNF) of muscle fatigue, and the Y-direction acceleration of the six-axis sensor were used as training data. In this study, a band-pass filtering technique was used to intercept and filter the sEMG and six-axis signals. Subsequently, a 10-second dataset was extracted at a sampling rate of 30 Hz for further analysis and processing. A total of 10,048 data sets were compiled and used as a database. We succeeded in training the decision tree (DT) at 93.56%, support vector machine (SVM) at 81.56%, random forest (RF) at 96.37%, K-nearest neighbor (KNN) at 89.65%, and Naive Bayes at 75.52% accuracy.