Algorithm for Human Fall Detection Based on Acceleration Measurement

According to the World Health Organization, a fall is defined as an unexpected event in which participant comes to rest on the ground, floor, or lower level. Falls are one of the most serious life-threatening events. Automatic detection of a fall can reduce the time of an arrival of medical attention and consequences of prolonged lying after a fall.In this paper, a novel algorithm is presented for a human fall detection based on acceleration measurement using the 3axis sensor placed in the pocket. This algorithm was tested on two data sets with simulated falls and various daily activities. The obtained results show that the proposed algorithm allows us to achieve both sensitivity of 93% and specificity of 94.5% at the same time. These are values much higher than currently reported in the literature.