Personalized Definition of Short Sleep Using Distributional Characteristics of Individual Long-Term Wearable Data

Background: Sleep duration recommendations currently rely on fixed, self-report–based thresholds, despite growing availability of wearable sleep data. We evaluated whether personalized definitions of short sleep derived from such data offer greater functional relevance.<br><br>Methods: A year-long observational study of 464 healthy adults yielded continuous wearable sleep tracking and daily ecological momentary assessments of wellbeing. Short sleep was defined using a fixed duration threshold of &lt;6h/night and personalized thresholds anchored to each individual’s sleep-duration distribution. Consecutive short-sleep nights were grouped into sequences to capture accumulative exposure. Linear mixed-effects models examined associations with short-term outcomes (alertness, perceived stress, sedentariness and autonomic markers) and longer-term cardiometabolic markers such as adiposity and arterial stiffness.<br><br>Findings: Personalized and fixed definitions of short sleep yielded similar average thresholds and overall prevalence of short-sleep nights and showed comparable associations with long-term cardiometabolic outcomes. However, personalized short sleep provided substantially better model fit and larger effect estimates for short-term outcomes, including alertness, sleep satisfaction, stress, resting heart rate, heart rate variability and sedentariness. Effects increased with successive short-sleep nights. Isolated short nights were common, while longer runs were relatively infrequent and terminated by socially constrained recovery sleep. Personalized thresholds distinguished stable short sleepers—who rarely crossed their individualized threshold—from individuals experiencing recurrent sleep deficiency.<br><br>Interpretation: Personalized, distribution-based definitions better capture night-to-night deviations from habitual sleep that drive short-term functional vulnerability. These findings support the use of individualized sleep metrics derived from long-term wearable data to inform scalable, person-specific guidance without medicalizing isolated short-sleep episodes.