Monitoring Health By Detecting Drifts And Outliers In Patterns Of An Inhabitant In A Smart Home

The elderly, along with people with disabilities or chronic illness, are most often dependent on some kind of formal or informal care. They are forced to move to a place where they can be cared for. Automatic health monitoring allows them to maintain their independence and continue living at home longer by continuously providing key health and activity information to caregivers. In this thesis, we present a novel technique, called the Health Monitoring System (HMS), which is a data-driven automated monitoring system for detecting changes in the patterns of activities/inactivity, health data and the living environment. HMS classifies these changes as drifts and outliers. These changes reflect short and long term lifestyle trends as well as any sudden changes in the living patterns of the inhabitant. HMS uses domain knowledge to determine the importance of a change and reports them to the caregivers in an easy-to-understand format.