Sarcopenia Recognition System Combined with Electromyography and Gait Obtained by the Multiple Sensor Module and Deep Learning Algorithm

• Published in: Sensors and materials Vol. 34; no. 6; p. 2403
• Main Authors: Chen, I-Miao, Yeh, Pin-Yu, Chang, Ting-Chi, Hsieh, Ya-Chu, Chin, Chiun-Li
• Format: Journal Article
• Language: English
• Published: Tokyo MYU Scientific Publishing Division 01.01.2022
• Subjects: Algorithms; Deep learning; Electromyography; Hardware; Inertial sensing devices; Machine learning; Modules; Muscles; Physicians; Recognition; Sarcopenia; Sensors; Signs and symptoms; Software; Wearable technology
• ISSN: 0914-4935; 2435-0869; 2435-0869
• DOI: 10.18494/SAM3787

At present, many diseases can be predicted through data obtained by wearable sensors. The majority of these proposed wearable devices only use inertial sensors to obtain the walking motion signals of a subject. However, since the symptoms of sarcopenia are reflected in the changes in human muscles, we propose a sarcopenia recognition system, which consists of hardware and software. The hardware is composed of multiple sensor module (MSM), which is a wearable device used to collect the signals of electromyography and gait (EAG). The software is composed of biomedical and inertial sensors algorithm (Bodi algorithm) and leg health classification net (LCNet). The Bodi algorithm is used to calculate various gait indicators after predicting the risk of sarcopenia obtained by LCNet. The accuracy of LCNet is 94.41%, its precision is 91.58%, its specificity is 95.81%, and its sensitivity is 91.58%. In the future, we expect to use the proposed MSM to collect additional subjects' gait data and apply it to other disease predictions to assist physicians in disease diagnosis.