A Situation-Driven Framework for Relearning of Daily Living Activities in Smart Home Environments
Daily living activities (ADLs) are sine qua non for self-care and improved quality of life (QoL). Self-efficacy is major challenge for seniors with early-stage dementia (ED) when performing daily living activities. Early-stage dementia (ED) troubles cognitive functions and thus impacts an elderly person’s functioning initiative and performance of instrumental activities of daily living (IADLs) that requires sequence of steps to accomplish. As a result, older adults are predisposed to safety-critical situations with life-threatening consequences which are a cause for concern in ambient assisted living (AAL). A safety-critical situation is a state or event that potentially constitutes a risk with life-threatening injuries or accidents. To address this problem, a Situation-driven framework for relearning of daily living activities in smart home environment is proposed. The framework is composed of three (3) major units namely: a) Goal inference unit – leverages a deep learning model to infer human goal in a smart home, b) situation-context generator – a risk assessment and safety reasoning model for ADLs, and c) a recommendation unit – to support decision making of seniors in safety-critical situations. We validated our proposed model against ADLs dataset collected from smart home. The results obtained were promising.
Committee: Carl Chang (major professor), Jennifer Margrett, Simanta Mitra, Jin Tian, Ying Cai