Using Time Series Techniques to Forecast and Analyze Wake and Sleep Behavior

Smart home technologies provide numerous benefits for providing healthcare to individuals in a non-invasive manner. Our goal of this research is to use smart home technology to assistance people recovering from injuries or coping with disabilities to live independently. In this paper, we propose an algorithmic method, Behavior Forecasting (BF), to model and forecast both the wake and sleep behaviors that are exhibited by an individual. The BF method consists of (1) detecting wake/ sleep cycles, (2) defining numeric values that reflect sleep behavior and numeric values that reflect wake behavior, (3) forecasting the numeric wake and sleep values based on past behavior, (4) analyzing the effect of wake behavior on sleep by using wake behaviors when forecasting for the next sleep behavior observed, and vice versa, and (5) improving the performance of score prediction by using both past wake and past sleep scores. We evaluate the performance of our BF method with data collected from CASAS smart homes. We found that incorporating time series techniques such as a periodogram improves the detection of regular sleep and wake cycles. We also found that regardless of the utilized forecasting method, we can model wake behavior and sleep behavior with the minimum accuracy of 87%. These results suggest that we can effectively model wake and sleep behaviors in a smart environment. Furthermore, that future wake behavior can be determined from sleep behaviors and vice versa.