A machine-learning based approach to pre-impact fall detection with wearable devices : MOTION MONITORING USING SENSOR FUSION AND THE SUPPORT VECTOR MACHINE
Simon Johansson · KTH Publication Database DiVA (KTH Royal Institute of Technology) · 2015
Falling accidents represents a major threat and is considered as a major source of morbidity and mortality among the elderly. As a consequence of fall related injuries, three people dies in Sweden every day. Additional factors, such as fear of falling further impacts the quality of life for the elderly. Due to the demographic change, which results in an increasing amount of elderly in the population, the costrelated to fall accidents is increasing. In order to the reduce the cost, preventivemethods and tools are believed to be a feasible approach. This report is the resultof a conceptual study that presents the issues related to the development of an individualizedmotion monitoring system applicable to pre-impact fall detection. Thestrategy adopted for fall detection, is to learn the normal behaviour of the user inorder to recognize fall as an anomaly from activities of daily living. The results are based on the comparison between an individualized, and a generalized algorithm.The conclusion is that the suggested algorithm is applicable in pre-impact fall detection systems.