Non-ergodic behavioral understanding focusing on interaction with the elderly and everyday items

Application to behavior analysis of elderly people and comparison before and after the introduction of welfare equipment

Abstract

Caregivers need to understand the daily living activities of elderly people to design living environments that meet their diverse needs. This paper proposes a system that automatically summarizes the daily activities of elderly people without defining in advance the behaviors to be analyzed, based on a motion-based analysis focusing on interactions between the elderly people and objects in the living rooms. To evaluate the effectiveness of the developed system, it was applied to three months of behavioral data of five elderly people in their 80s residing in nursing homes. The experimental results demonstrate that the system can detect abnormal behaviors such as repeated unnatural drawer use. In addition, this report presents that the system can detect behavioral changes due to installing new furniture in the environment by comparing daily activities.