O062 Sleep behaviour modification in shift workers: Applying models of arousal dynamics tailored to individual shift schedules and personal commitments

INTRODUCTION: Circadian-based biomathematical models of sleep-wake and alertness regulation can improve sleep health in shift workers by providing physiologically optimal times for sleep. It is unclear whether they can be successfully implemented in real-world contexts. This study, for the first time examined associations between model-based sleep-wake recommendations and modification of sleep behaviours in shift workers. METHODS: Nurses (n=28; 37.2±9.6yrs) with matching-rotating shift schedules received personalised sleep recommendations from either the Phillips-Robinson Model or Model of Arousal Dynamics. Nurses used sleep diaries to record sleep-wake behaviours: (i) pre-exposure to recommendations (baseline, 1-week), (ii) upon exposure (1-week) and completed PROMIS-sleep disturbance, sleep-related impairments pre and post. RESULTS: During baseline, participants usual bedtimes and waketimes were at least 180mins away from recommended times from both models. Upon exposure to recommendations, sleep behaviours were aligned within: (i) 62±11mins for bedtime, 71±9mins for waketime (Model of Arousal Dynamics), (ii) 104±16mins for bedtime, 89±12mins for waketime (Phillips-Robinson model) with large effect sizes (p<.00001 for both models). Participants sleep windows had better compliance with recommended sleep windows for the CRC Alertness model (79% overlap) compared to the Phillips-Robinson model (69% overlap). Sleep disturbance and sleep-related impairments improved significantly upon study conclusion (small-effect sizes both groups). Participants reported that automated recommendation delivery, and use of diurnal preference and real-time sleep behaviours to further personalise recommendations will facilitate greater uptake. CONCLUSION: The study provides positive proof-of-concept for the implementation of biomathematical models of sleep-wake dynamics in shift work contexts. Accounting for individual differences may improve applicability of recommendations.