Towards Person-Centered Anomaly Detection and Support System for Home Dementia Care

Anomaly detection is a crucial issue for people with dementia and their families to live a safe and comfortable life at home. The elderly monitoring system is a promising solution. However, the conventional systems have limitations in detectable anomalies and support actions, which cannot fully cover individual needs. To achieve more person-centered home care for people with dementia, our research group has been studying environmental sensing with IoT. In this paper, using the environmental sensing, we propose a new service that allows individual users to customize definition of anomaly and corresponding actions. Specifically, borrowing a mechanism of context-aware services, we regard every anomaly observed within the house as a context. We then define every care as an action bound to an anomaly context. This achieves the personalized anomaly detection and care. To demonstrate the feasibility, we implement a prototype system and conduct a practical case study.