Bloomfield · PLOS digital health 2024 · prospective longitudinal cohort study · n=525

Predicting stress in first-year college students using sleep data from wearable devices.

Cited 35 times in the scientific literature.

Level 3 - non-randomized controlled study

Prospective longitudinal observational cohort study.

PubMed 38602898 · doi:10.1371/journal.pdig.0000473 · record verified 2026-08-29

What was done

Participants in the Lived Experiences Measured Using Rings Study (LEMURS), consisting of 525 first-year public university students, wore smart rings providing continuous biometric data and completed weekly surveys between October and December 2022. Mixed-effects regression models examined weekly associations between wearable-derived sleep metrics and self-reported perceived stress, adjusting for gender and week of the semester.

What was found

Wearable sleep and autonomic metrics were significantly associated with self-reported stress: - Total sleep time (TST): each additional hour reduced odds of moderate-to-high stress by 38.3% (OR = 0.617, p < 0.01). - Resting heart rate (RHR): each 1 bpm increase raised odds of moderate-to-high stress by 3.6% (OR = 1.036, p < 0.01). - Heart rate variability (HRV): each 1 ms increase decreased odds of moderate-to-high stress by 1.2% (OR = 0.988, p < 0.05). - Average respiratory rate (ARR): each 1 breath/min increase raised odds of moderate-to-high stress by 23.0% (OR = 1.230, p < 0.01). Participants who did not identify as male (female, nonbinary, and transgender) and later weeks in the semester were also significantly associated with higher self-reported stress.

Why it matters

Demonstrates that consumer wearable sleep and physiologic metrics (duration, autonomic tone, respiration) track changes in subjective stress levels over time in college students, supporting the potential of passive biometric sensing for digital mental health monitoring.

Limits

The study was conducted at a single public university over a single academic quarter (roughly 3 months) in first-year undergraduates, limiting generalizability. Sleep parameters relied on proprietary wearable estimates rather than polysomnography, and perceived stress was assessed via self-report surveys rather than clinical psychiatric evaluations. Observational design precludes causal inference.

Cited by