Daily Activities Forecasting for Long-Term Elderly Behavior Change Detection

Abstract

The increasing availability of sensors and intelligent objects enables the design of new healthcare applications and services for elderly people. Such applications exploiting smart homes can be designed to improve the daily lives of older people, allowing them to stay at home longer, while receiving the health care they need. In this article, we present an original solution to conduct a long-term analysis of elderly’s behavior in order to detect slow changes, which may traduce the early stage of a disease. The originality of this work is to use diverse contextual factors, including weather, holidays, and seasons, to learn dynamic routine behavior patterns in aging individuals through a clustering method iteratively adapts via a moving-window mechanism, and then exploit machine learning techniques on previously observed Activities of Daily Living (ADL) to forecast when the future ones should occur. We use this forecast to measure the actual gap with the future ADLs, as soon as they occur, and detect behavior changes of the monitored elderly people and thus specify its type (degradation or recovery). Experimental results are proposed to show the superiority of the proposed method compared to the existing state-of-the-art methods.