A new adaptive algorithm for detecting falls through mobile devices

Abstract

Most of elderly people suffer physical degeneration that makes them particularly vulnerable to falls. Falls cause injuries, time of hospitalization, re-habilitation which is particularly difficult for the elderly and disabled. This paper presents a new system with advanced capacities for learning and adaptation specifically designed to detect falls through mobile devices. The systems proposes a new adaptive algorithm able to learn, classify and identify falls from data obtained by mobile devices and user profile. The system is based on machine learning and data classification using decision trees. The main contribution of the proposed system is the use of posturographic data and medical patterns as a knowledge base, which notably improves the classification process. © 2011 Springer-Verlag Berlin Heidelberg.