Feature fusion for sleep bruxism detection and prevention based on micro motion accelerations from MEMS sensor and EMG, ECG and HRV signals analysis.

A system of synchronous recording and analysis of both direct motion signals acquired from MEMS accelerometer and physiological EMG, ECG, Heart Rate Variability (HRV) collected from masseter muscle is presented as an on-line tool for early sleep bruxism (SB) episode detection and prevention. A hybrid feature set extracted on-line after time-frequency analysis of multi-sources data is the input for final decision rules classifier. Results regarding sensitivity and specificity of 20 supervised trials of classifying extracted feature fusion, validated the significant meaning of micro-motion accelerations recorded from MEMs sensors, added to the system for SB episodes detection quality improvement.