Grant: Discovery Grants Program - Individual

The rising popularity of virtual care programs in the Canadian health system has amplified the demand for practical, safe and innovative eHealth patient monitoring solutions. The emergence of smart homes, outfitted with sensors and IoT devices, offer an innovative eHealth solution to monitor and support patients to perform self-care Activities of Daily Living (ADL) at their homes. Activity Recognition (AR) is an Artificial Intelligence (AI) driven approach to autonomously recognize the activities of an individual using low-cost sensor data; by not using video monitoring AR is a privacy-preserving method for patient monitoring. Combining smart homes with AI-based AR offers a novel approach to eHealth patient monitoring, whereby we can monitor an patient's self-care ADL, and help them complete the self-care ADL by providing personalized guidance and behaviour change programs to instil self-sufficiency to help him/her complete the prescribed self-care ADL. The long-term objective of the proposed research program is to investigate and develop AI-based data- and knowledge-driven AR methods to implement a privacy-preserving eHealth patient monitoring platform tp deliver virtual care to individuals with chronic diseases, dementia, disability and the elderly. The short-term objective is develop knowledge-driven AR solutions for eHealth based patient monitoring by novel semantic web based knowledge-driven AR methods that (a) improve the accuracy, robustness and flexibility of AR methods for recognizing self-care ADL within a smart home, and (b) demonstrate a prototype AR based eHealth patient monitoring platform that offers (i) personalized self-care ADL assistive acts; and (ii) behaviour change interventions to help patients perform self-care ADL. The societal objective is to improve the wellbeing of Canadians, whilst saving cost to the Canadian health system. This research is inspired by Artificial Intelligence (AI) in Medicine. T he research objectives will be pursued by developing Knowledge-driven AR methods to (a) Represent ADL processes using semantic models to achieve AR robustness and flexibility; (b) Represent an AAL environment using semantic models, to achieve AR accuracy and flexibility; and (c) Recognize self-care ADL, using semantic reasoning, whilst handling any uncertainty due to sensor noise and incorrect task sequences, to achieve AR accuracy and robustness. The research results will be translated as a prototype AR based eHealth patient monitoring platform. The research will impact eHealth based virtual care by demonstrating the technical feasibility of low-cost privacy-preserving AR methods for remote patient monitoring. The AR methods will be applicable for real-world applications, including astronaut monitoring, smart classrooms and security. The research will advance autonomous eHealth based virtual care applications, such as the use of personal care robots and IoT devices to provide timely assistance to vulnerable individuals.