Conversational AI for Monitoring Behaviour and Well‐being of People Living with Dementia

Abstract

AbstractBackgroundConversational AI offers an intuitive and natural user interface with significant potential to promote the independence, well‐being, and safety of people living with dementia (PLWD) at home. This technology, such as smart speakers and social robots, can be used to access information, set reminders, assist with activities of daily living (ADL), stimulate communication, encourage behaviour, or alert the caregiver. However, voice technology has yet to demonstrate a direct impact on supporting PLWD in daily living contexts, particularly with health and well‐being monitoring utility. Adaptation to individual changing needs and cognitive abilities for personalised support remains a major research challenge. Further studies are required to assess the impact of conversational technology in different cultural and economic contexts.MethodWe investigate conversational social robots that engage with PLWD to query health and well‐being, map household behavioural patterns, and assist with ADL. We analysed in‐home activity data, including regular interactions with Amazon Alexa, from 14 households with PLWD (75‐94 years, four females) over 388 days. We examined the overall usage of conversational technology over time. We further considered activity sequences preceding or following Alexa interactions to trace behavioural patterns.ResultA significant decrease in conversational technology usage was observed after the novelty phase (i.e., >3 months) across the cohort. We argue this overall decline in user engagement is due to the lack of perceived utility and adaptation of interactions over time. Initial findings from our data mining approach identified behavioural patterns in household activity before and after user‐initiated interactions with Alexa. Additionally, ongoing studies are exploring the feasibility of implementation in both the UK and India with a specific focus on the impact of conversational AI in dementia care and the challenges associated with different cultural settings.ConclusionIf well incorporated into the living environment, conversational technology can monitor behaviour and well‐being of PLWD. This may be accomplished by proactively verifying issues detected from other sensors in smart environments and alerting caregivers for appropriate intervention. An early focus must be placed on personalising interactions, ensuring accessibility and cultural feasibility.