Wearable Hybrid Strain-Myoelectric Sensing System for Machine-Learning-Assisted Sarcopenia Screening

The early screening of sarcopenia represents a critical clinical need amid the accelerating global aging population. Current diagnostic methods, relying on bioelectrical impedance analysis (BIA), handgrip strength testing, and other clinical examinations, depend on costly medical equipment and struggle to concurrently assess both muscle mass and strength. Herein, we propose a Wearable Sarcopenia Assessment System (WSAS), which employs an integrated hybrid surface electromyography (sEMG)-piezoelectric strain sensing platform to synchronously capture electrophysiological signals and mechanical deformation signals during muscle contraction in handgrip tests (signal-to-noise ratio: 34.32 dB), and incorporates a CNN-LSTM deep learning framework. This model was trained using nine physiologically relevant features (including root mean square (RMS), mean absolute value (MAV), and integrated EMG (iEMG)) extracted through feature engineering as prior knowledge. Validated in a cohort of 75 elderly participants, the proposed system achieved a screening accuracy of 99.85% with an area under the curve (AUC) of 0.97. Shapley additive explanations (SHAP)-based interpretability analysis further revealed that WSAS captures neuromuscular alterations associated with sarcopenia, including type II-to-type I muscle fiber transition and neuromuscular junction remodeling. These results demonstrate the potential of WSAS as a portable, low-cost, and radiation-free platform for early-stage sarcopenia screening.