In order to support ageing in place for elderly people, technologies and services for home environments need to be developed. An intervention mechanism is proposed in this paper in a smart home environment to provide reminders to assist elderly inhabitants to complete activities of daily living (ADL). The situation of multiple inhabitants in a single smart environment is addressed. A probabilistic learning approach is proposed to characterise inhabitants' behavioural patterns, learned from summary activities collected during a period. Activity reasoning can then be carried out given partially observed low-level sensor information. Decision support is used to monitor inhabitants' activities and thus to assist the completion of tasks if necessary. Personalised reminders at various levels of detail can be delivered based on individual need and preference. Appropriate thresholds are learned to be used to ensure delivery of predictions for which confidence is high, to avoid confusing inhabitants with incorrect reminders. The potential of our approach to support assistive living and home-health monitoring of elder patients is demonstrated.
|Journal||Journal of Ambient Intelligence and Smart Environments|
|Publication status||Published - 2010|