Longitudinal Infant Sleep Monitoring Using a Sensor-Enabled Responsive Bassinet: A Population-Scale Feasibility Study

Sleep is crucial to infant development, and excessive sleep disturbances are associated with adverse outcomes for both infants and their caregivers. There is limited information on the longitudinal development of sleep (e.g., duration, fragmentation, etc.) from birth to 6 months of age. New technologies, which include real-time environmental sensing and responses, have the potential to overcome many of the traditional limitations on infant sleep monitoring. In this study, we demonstrate the feasibility of utilizing aggregated activity logs from a commercially available IoT (Internet of Things) bassinet to derive traditional sleep metrics (longest sleep stretch, total night sleep, and sleep efficiency), as well as novel metrics related to infant fussing and impacts of the bed's ability to deliver responsive motion and sound. A total of 26,187 infants (1000-8000 per night) were included in this analysis. A data-driven approach was utilized to define the temporal boundaries of each night, divide each night into periods of sleep and fussing, and identify appropriate nights for inclusion. The derived data provide, in unprecedented resolution, a detailed longitudinal view of infant sleep in this specific population. Our results generally align with previous studies of traditional sleep metrics; however, they also demonstrate a methodological framework for descriptive or comparative monitoring of sleep and soothing, and uniquely characterize dyadic interactions that are not well-captured by traditional metrics. For example, the bassinet's activity logs indicate not only the proportion of fussing episodes that are resolved without caregiver intervention (e.g., removal), but also reflect the delay between fussing and the need for caregiver intervention. Further evaluation of this sensor-enabled, responsive technology in relation to sleep and fussing is merited.