Supervised machine learning scheme for wearable accelerometer-based fall detector

Falling down is one of the main causes of trauma, disability and death among older people. Inertial sensors and accelerometer-based devices are able to detect falls in controlled environments. The aim of this work is the development of a computationally low-cost algorithm for feature extraction and the implementation of a machine learning scheme for detection of fall events in the elderly, by using the 3-axial MEMS wearable wireless accelerometer. The proposed approach allows to generalize the detection of fall events in several practical conditions, after a short period of calibration. It appears invariant to age, weight, height of people and relative positioning area (even in the upper part of the waist), overcoming the drawbacks of well-known threshold-based approaches in which several parameters need to be manually estimated according to the specific features of the end user. The supervised clustering step is achieved by implementing a One Class Support Vector Machine (OC-SVM) classifier in a stand-alone PC. A polynomial kernel function is used in order to limit the computational workload while maintaining high performances in terms of reliability and efficiency