Development of a Wearable IMU-based Fall Detection System for the Elderly

This paper presents the development of a wearable fall detection system for the elderly using a fixed threshold-based algorithm. The system automatically detects the fall occurrence and sends real-time alert message with GPS coordinates to the emergency contact. Manually, the user can also press a pushbutton when a fall event is anticipated. The integration of GPS module allows the geofencing features into the system as an alert shall the elderly moves beyond the safe pre-defined zone. Three sensors are incorporated into the system namely accelerometer, gyroscope and magnetometer to determine the motion, angular velocity and orientation. Experimental trials demonstrate a good accuracy of 86.67% under real-time environment indicating the system reliability and responsiveness. The proposed solution is cost effective with less complex design and suitable for the use of continuous monitoring application particularly for the elderly at risk of falling including those with dementia or Alzheimer's disease.