Beyond the Circadian Rhythm: Variable Cycles of Regularity Found in Long-Term Sleep Tracking

Sleep is more than resting eight hours a day-it contextualizes and shapes the routines during the day. Using a large-scale naturalistic dataset of 180,083 people from a popular sleep app, made possible by the widespread adoption of passive tracking, we find that people's lives have distinct natural rhythms that can be automatically inferred from sleep routines. We discover heterogeneous behaviors: the rhythm of sleep is different for each person, as there is a different cadence for each person to achieve consistency. Some are most consistent week-to-week, while others weeks-to-weeks. We investigate changes in overall daily routines and find the interval for each person at which they show the most consistency. Through a series of comparative case analyses, we investigate the implications of designing for the weekly 'norm'. Our tripartite analyses triangulate to one conclusion: we should design for people's natural routines to account for variable cycles of regularity.

• Human-centered computing → Empirical studies in HCI; Empirical studies in ubiquitous and mobile computing; • Applied computing → Consumer health.