Recently, a number of projects have been undertaken to collect continuous behavioral data from elderly individuals using unobtrusive in-home sensors. An important challenge facing these projects is the development of approaches for interpreting these data. One approach, based on statistical process control, is to model each individual's behavior as a random process whose mean and variance may change over time. The sensor data then provide specific measures of the process that can be used to identify changes in patterns of behavior. The approach is presented and applied to measures of sleep hygiene in a group of 14 community-dwelling elders monitored over a six month period. Both acute and slow changes in the patterns of sleep were successfully identified in individuals using this approach.