Responsive Home - Integrating Machine Learning with Home-based Sensors to Provide Incident Response for Seniors Living Independently

点击翻译按钮获取中文摘要

Like many other industrialized nations, Canada faces the challenge of adapting to a rapidly aging demographic.Our country requires innovative and transformative solutions in order for our health care and social systems toremain strong and sustainable, in presence of a growing population of senior citizens. The ability to "age inplace" and remain living independently for a longer period of time is an important objective in this context.Most Canadians prefer aging in place and independent living over other forms of care (e.g., residential care), ifthey feel adequately supported and cared for. Information and communication technologies (ICT) are expectedto play a major role in enabling this vision. A senior's home environment equipped with smart sensors may beable to prevent or detect health and safety-related incidents, such as falls or the onset of an illness.Since there is a large diversity of lifestyles and living environments, such "Smart Home" technologies mustbe adaptive to the particular habits and needs of an individual. Machine learning algorithms can be used toprovide such intelligent adaptation based on activity data streams sourced from a variety of home-basedsensors. Our company partner (Tochtech) is a healthcare technology provider start-up firm who is developing aSmart Home package in support of caring for independently living seniors. Tochtech's solution will be capableof monitoring daily activity data of seniors and raising alerts when detecting potential safety or health risks.Tochtech is interested in exploring the use and evaluation of machine-learning algorithms, platforms and toolsfor the purpose of providing an adaptive incident detection system. They have engaged with the University ofVictoria's LEAD lab to provide expertise and guidance. The objectives of this NSERC Engage project are (1)to evaluate current machine learning algorithms, platforms and tools and select the ones most suitable for theapplication problem, (2) to prototype a machine learning system for detecting two selected risk categories(accidental falls and acute illness), and (3) to design an extensible architecture platform for adding detectionalgorithms to address additional risk categories.