A Novel Double Pressure Sensors-Based Monitoring and Alarming System for Fall Detection

Considering non-wearable devices or systems, two technologies, camera-based and sensor-based, are frequently investigated and implemented. Most people who concern very seriously about their privacy choose the sensor-based one. Previously, a pressure sensor-based system called “NEF” was developed to monitor on-bed movements and send out alarming notifications to end-users when undesired events occurred. The system used only one pressure sensor that was placed on the bed or mattress. On-bed positions and patterns were classified with high accuracy. However, the fall event is actually very dynamic. Movements outside the bed shall be included. In this paper, a novel double pressure sensors-based system was proposed. Applying different machine learning techniques, random forest yielded the best fall detection model with 100% accuracy. Further experiments in multiple centers will be investigated.