Identification of Neurological Markers of Sarcopenia Disease Using Functional Near-Infrared Spectroscopy and Machine Learning

Abstract

Sarcopenia, a disease defined by the loss of muscle mass and function, plays a significant role in the quality of life of the elderly. Recent studies suggest that the loss of muscle strength and function associated with sarcopenia may be linked to neural control mechanisms. This study aimed to find a neuro-cognitive biomarker for sarcopenia and to classify it using fNIRS and machine learning methods. Connectivity matrices created from fNIRS data obtained from the Hand Grip experiment, conducted on 50 participants (27 controls, 23 sarcopenic), were used as features in the classification. This resulted in the Linear SVM model showing the highest performance with an 87.4% accuracy rate and 0.94 AUC value. These results indicate that functional connectivity data obtained through fNIRS could serve as an objective biomarker for sarcopenia classification, and that high-performance classification is feasible using this biomarker.