부산대학교 도서관

There are an increasing number of elderly people with various pathologies and care needs, making nursing homes one of the most suitable places for them to live. The demand for nursing homes is expected to increase further in the coming years due to the growing life expectancy. Human activity recognition (HAR) approaches are an excellent resource that can be deployed in nursing homes to detect an inhabitant’s activity through sensors in the environment. Despite the many theoretical approaches to HAR proposed in the literature, there is a significant lack of implementation in real environments. In this paper, we present an activity recognition approach deployed to monitor a floor of a nursing home using a mobile app called ACTIVA, which allows caregivers to track inhabitants and be notified of anomalies in real-time. The ACTIVA system relies on successful tools proposed for HAR, such as the use of fuzzy logic techniques, linguistic protoforms, and temporal windows to infer the location and activity of each inhabitant in a multi-occupancy environment. These tools have proven effective in inferring inhabitant activity and location, allowing for more precise and detailed monitoring. ACTIVA provides multiple benefits in terms of providing more information on the activities performed by residents, reducing stress for caregivers, and increasing safety for the residents being monitored. To evaluate the impact of ACTIVA app, we gathered feedback from different stakeholders and found overall positive satisfaction levels. Our approach represents a significant step towards the practical implementation of HAR approaches in real nursing home environments.