Abstract Introduction Lifestyle behaviors may influence sleep architecture and quality. The extent to which sleep parameters are impacted by day-to-day behavioral variations remains poorly understood in large, real-world populations. Methods Sleep was monitored from 11/15/2023-11/14/2025 using a commercially available under-mattress device (Sleeptracker-AI Monitor, Fullpower Technologies Inc., California, USA) that continuously records sleep via piezo-electric sensors. Sleep metrics included wake after sleep onset (WASO), total sleep time (TST), REM and deep sleep percentages, breathing anomaly index (BAI), and arousal index (AI). Each morning, participants self-reported exercise, alcohol consumption, eating, sleep-aid use, and TV or mobile device use. For each question, only participants who provided at least two distinct responses across multiple nights were included to enable within-participant comparisons. Mixed linear models with participant as a random effect were used to derive percentage changes in sleep parameters associated with self-reported behaviors. Results Analyses varied in sample size (largest exercise: 85,706 participants, 2,318,639 nights; smallest sleep-aid use: 2,440 participants, 88,966 nights; mean across all analyses: 37,304.4 participants, 997,703.6 nights). Across all participants (48.2% female), the mean age was 52.9±14.4 years. All reported effects were significant (p 0.001). On nights following exercise (vs. no exercise), participants had reduced WASO (some: -1.60%; vigorous: -1.63%), reduced BAI (some: -0.99%; vigorous: -2.44%), increased deep sleep percentage (some: +0.50%; vigorous: +0.63%), and reduced REM sleep percentage with vigorous exercise only (-0.67%). Alcohol use (vs. non-use) was associated with increased WASO (+3.97%), reduced REM sleep percentage (-1.71%), increased deep sleep percentage (+1.38%), reduced BAI (-2.51%), and reduced AI (-1.26%). Eating within 2 hours of bedtime (vs. not eating) increased deep sleep percentage (snack: +0.50%; meal: +1.45%), reduced BAI (snack: -1.07%; meal: -1.01%) and TST (snack: -0.51%; meal: -0.46%), and had opposing effects on WASO (snack: -0.83%; meal: +0.72%). Compared to non-use, sleep-aid use reduced WASO (-4.01%), reduced REM sleep percentage (-1.33%), and increased TST (+2.67%). TV or mobile device use (vs. non-use) was associated with decreased WASO (-1.68%) and TST (-1.10%). Conclusion Lifestyle behaviors were associated with within-person differences in objective sleep metrics. Non-invasive sleep-tracking technologies offer a powerful way to characterize how lifestyle choices influence sleep health. Support (if any) Fullpower Technologies.
Kamaci et al. (Fri,) studied this question.