Remote Monitoring System for the Elderly people based on a Contextual Bandit approach

Alzheimer's disease poses a significant challenge for patients and their caregivers, necessitating innovative solutions to enhance patients quality of life and safety. In this article, we propose an assistance system based on a Contextual Bandit (CB) approach to help elderly Alzheimer's patients in performing their activities of daily living autonomously. Our system relies on a human activity recognition system and a prompt system to alert the patient when needed. For decision-making, we employ a Contextual Bandit Reinforcement Learning module capable of identifying and suggesting necessary assistance based on the patient's behavior. We validated the effectiveness of our system by testing it on the DemCare dataset. The results demonstrate the system's ability to provide tailored and responsive assistance to the needs of Alzheimer's patients.