Multi-Agent System for Detecting Elderly People Falls through Mobile Devices

Abstract

Falls in the elderly and disabled people represent a major health problem in terms of primary care costs facing the public and private systems. This paper presents a multi-agent system capable of detecting falls through sensors in a mobile device and act accordingly at runtime. The new system incorporates a fall detection algorithm based on machine learning and data classification using decision trees. The base of the system are three types of interrelated agents that coordinate to know the position of a user from data obtained through a mobile terminal, and GPS position, which in case of fall may be sent via SMS or by an automatic call. The proposed system is self-adaptive, since as new fall date is incorporated, the decision mechanisms are automatically updated and personalized taking into account the user profile.