Research Project
6832438
[unreadable] DESCRIPTION (provided by applicant): The research objective of the proposed work is to fuse sensor data to infer behavior of occupants in a home setting. The sensor data have been collected by CareWheels Corporation, a Section 501 (c)(3) non-profit public benefit corporation. The CareWheels research project, "lnternet-enabled Assistive Technologies for Independent Living and Aging-in-Place" has implemented a home sensor testbed currently operating at Pine Point Apartments, an independent living facility for people with severe physical disabilities in Portland, Oregon. CleverSet will employ its relational probabilistic modeling software environment to cluster behavior in unlabeled sensor data using unsupervised learning, and to discover behavioral models in labeled sensor data supplemented with simulated events, using supervised learning. CleverSet will also evaluate the models by using (1) expert knowledge and (2) a quantitative assessment of the models, including evaluation of models by mutual information, log probability scores and ROC curve analysis. [unreadable] [unreadable] The commercial impact of the research will be software capable of driving a new generation of home sensors capable of inferring the behavior of elderly occupants. The proposed technology will be a boon to frail elders who desire to live in their own homes as long as possible and to caregivers who are supervising elderly loved ones with dementia. In tandem with existing technologies, the proposed work will provide a level of respite from the hypervigilant stance necessary to monitor many dementia-afflicted elders. [unreadable] [unreadable]
