Health Risk Assessment and Monitoring System Based on Multi-sensor Information Fusion and Contrast Learning

Abstract

The number of solitaries is on the rise worldwide. In order to improve the risk resistance ability of loners to unexpected events, and to evaluate their life and health status in real-time and automatically warn their risks, a health risk assessment and monitoring system based on multi-sensor information fusion based on comparative self-supervised learning are proposed. The scheme predicts the vital signs and safety level of the solitary person by observing and calculating the preset multi-modal sensor information, integrates the information, identifies the routine, risk, emergency, and other states. The model was trained, verified, and tested by the physical signs data of 143 anonymous volunteers. The results show that the health risk assessment and monitoring system can effectively filter the errors of sensor information data, and the steady-state error is less than 5%, which has higher accuracy and efficiency than other recent information fusion methods. The model trained with sufficient historical data can significantly improve the ability of loners to cope with unexpected events.