A skin-conformal liquid metal strain sensor integrating deep learning for neck motion monitoring

Neck joints are susceptible to injury because of their extensive mobility and complex motion patterns, motivating the development of wearable motion sensors for health monitoring. Herein, we design a skin-conformal motion-monitoring strain sensor (SMS) based on a semi-liquid metal electrode and an adhesive encapsulation layer that promotes robust skin attachment. The sensor, fabricated by a simple printing method, exhibits consistent signal output across a broad range of stretching speeds, linear response up to 60% strain, detection of subtle 1% strain, and durability over 2,000 cycles without significant drift. Its strong adhesion also enables reusability and water resistance. When applied to multi-site neck motion recording and coupled with a convolutional neural network, the system classifies diverse neck movements with 96.9% accuracy. This work illustrates the promise of soft conductive materials and machine learning for practical neck healthcare applications.