Fall detection algorithm based on triaxial accelerometer data

Elderly people who experience a fall usually need urgent medical attention, as the consequences can be very serious. The response time in case of a fall can be greatly improved with the help of an automatic fall detection system that could raise an alarm whenever a fall is detected. This paper presents a threshold-based fall detection algorithm that processes data from a triaxial accelerometer in order to detect falls. The algorithm was designed to be implemented in a mobile system that uses a microcontroller for data processing and a triaxial accelerometer for data acquisition. In terms of performance, sensitivity of 97.05% and specificity of 99% were achieved on a data set with 34 simulated falls and 200 daily activities.