Detecting Health and Behavior Change by Analyzing Smart Home Sensor Data

Abstract

Smart home environments offer an unprecedented opportunity to unobtrusively monitor human behavior. Sensor data collected from smart homes can be labeled using activity recognition to help determine whether relationships exist between behavior in the home and health changes. To detect and analyze behavior changes that accompany health events, we introduce the behavior change detection (BCD) approach. BCD detects activity timing and duration changes between windows of time, determines the significance of the detected changes, and analyzes the nature of the changes. We demonstrate our approach using two case studies for older adults living in smart homes who experienced major health events, including cancer treatment and insomnia. Our algorithm detects behavior changes consistent with the medical literature for these cases. The results suggest the changes can be automatically detected using BCD. The proposed smart home, activity recognition algorithms, and change detection approach are useful data mining techniques for understanding the behavioral effects of major health conditions.