"Utilization of smart thermostat data at the population level to identify sleep parameters and time spent at home" (Preprint)
BACKGROUND Sleep behaviour and time spent at home are important determinants of human health. Research on sleep patterns has traditionally relied on self-reported data. This methodology suffers from bias and population-level data collection is challenging. Advances in Smart Home technology and the Internet of Things (IoT) have the potential to overcome these challenges to behavioural monitoring. OBJECTIVE The objective of this study is to evaluate the use of smart home thermostat data to evaluate household sleep patterns and the time spent at home, and how these behaviours are influenced by weekday, seasonal and seasonal weekday variations. METHODS The 2018 ecobee "Donate your Data" dataset for 481 North American households was collected for use in this study. Daily sleep cycles were identified based on sensor activation and used to quantify sleep time, wake-up time, sleep duration, and time spent at home. Each household's record was divided into different subsets based on seasonal, weekday, and seasonal weekday scales. RESULTS Overall, our results indicate that sleep parameters (sleep time, wake-up time, and sleep duration) were significantly influenced by the day of the week but were not strongly affected by season. In contrast, time spent at home was dependent on both weekdays and the season. CONCLUSIONS This is the first study to utilize smart home thermostat data to monitor sleep parameters and time spent at home and their dependence on weekdays, seasonal, and seasonal weekday variations at the population level. This type of analysis can influence and report on public health policy at the population level.