Fall Detection by Recognizing Patterns in Direction Changes of Constraining Forces: Tagungsband der eHealth2012

The severe consequences accidental falls may cause for elderly have pushed the development of mobility safeguarding systems. Threshold based fall detection algorithms in smartphones have shortcomings as accelerometer values from different devices are hardly comparable. In this work, a pattern-recognition procedure is presented whose main input consists of the change of direction of the constraining force that is exerted on an accelerometer attached laterally on the hip area. The algorithm makes use of fast wavelet decomposition and support vector machine theory.