Pre-bed and nighttime screen use, beyond daily total, is inversely associated with sleep quality: a longitudinal study of 350,600 nights

Sleep is fundamental to metabolic regulation, cognitive performance, immune function, and cardiovascular health, and evening screen exposure is widely proposed as a behavioural contributor whose adult evidence base remains thin. Here we analyze 350,600 paired screen-day and following-night observations from 3,086 Ultrahuman Ring AIR adult users. Sleep quality was assessed via the rings composite sleep score, derived from heart-rate variability, nightly movement, and skin temperature. At the user level, users in the highest screen-time quintile had lower sleep scores (Cohens d = -0.30), shorter sleep duration (d = -0.25), and lower sleep efficiency (d = -0.14) than the lowest screen-time quintile (all q [&le;] 0.005). Further, 45+ min of screen use in the last hour before bed was associated with mean sleep scores at the bottom of the cohort range, whereas the same dose 4-5 hours earlier showed no detectable cost, so the timing of screen use, not just its total, mattered. We next asked whether the way users distribute their screen time across the 24 hours, independent of total dose, separates users by sleep outcome. K-means clustering on 24-hour screen-use profiles identified three phenotypes: Daytime Peakers (DP), Late-Night Users (LNU), and Round-the-Clock Users (RCU), distinguished primarily by their nighttime share of 24-h screen use (DP 8.2%, LNU 16.9%, RCU 29.3%). Despite comparable total daily screen time, the phenotype gap in mean sleep score between DP (75.2 {+/-} 0.3 SEM) and RCU (66.7 {+/-} 0.6) was 8.5 points. We further identified users who transitioned phenotypes across four sequential quarters of follow-up; in this longitudinal subcohort, the magnitude of sleep-score change tracked the magnitude of the behavioural shift, with DP [-&gt;] LNU transitioners declining by 5.16 {+/-} 0.94 points and LNU [-&gt;] RCU transitioners by 4.79 {+/-} 1.87 points (both p < 0.05). Together, these findings position the temporal distribution of screen use, alongside its daily total and its concentration immediately before bed, as a behavioural correlate of objectively measured sleep quality in adults.