Development and Experimental Evaluation of Activity Detection System using Single Point Sensor for Observing Daily-Life of Elderly People

A new support system is proposed for monitoring and advising elderly people. The proposed system uses a single microcomputer with various sensors to collect data in a room of the house where the elderly is living alone. It detects a gradual change of activity of the elderly by extracting general pattern of the activity from the measurement data (especially, time series data of sound strength) of the sensors. When the activity patterm markedly changes from the general model, notifications and advice are sent to not only the elderly but also their family with the aim of maintaining the good physical condition of the elderly and encouraging them to perform health-aware activities. As a first step of the study, the feasibility of behavior estimation from the collected sound data was evaluated. The evaluation results clarified that the measured sound data can be used to estimate the behavior of elderly people.