0815 A Machine Learning Approach to Identify Factors Associated with Adolescent Sleep Outcomes
Emily Ricketts,Katherine Kaplan,Dana McMakin,Guadalupe Patriarca,Gina Mathew,Tylor McGrew,Anne-Marie Chang,Lauren Hale
DOI: https://doi.org/10.1093/sleep/zsae067.0815
IF: 6.313
2024-04-20
SLEEP
Abstract:Abstract Introduction Beyond natural physiological changes coinciding with puberty, several factors perpetuate poor sleep health in adolescents. However, there is limited research examining their relative importance. This investigation used two popular machine learning approaches to identify the relative significance of key factors associated with multiple sleep outcomes, including bedtime, sleep duration, and social jetlag, in a diverse sample of adolescents. Methods Participants were 3,381 adolescents from the age 15 wave (M=15.59, SD=0.76) of the Future of Families and Child Wellbeing Study. Sixty-eight factors spanning child- and parent-reported socio-demographics, neighborhood environment, sleep behaviors, activities, psychopathology, family, school, physical health, and dietary patterns, were identified through literature review and author consensus. Variables were entered into lasso penalized regression (LASSO) and random forest machine learning models for variable selection and establishing the order of factor importance. Self-reported social jetlag (sleep midpoint on weekend nights minus sleep midpoint on school nights) and weekday and weekend bedtime and time in bed were dependent variables. Results Factors overlapping between LASSO and random forest in their association with sleep outcomes are reported. Delinquent behavior, peer delinquency, consuming sweet drinks, parent removal of adolescent privileges, and smartphone use near bedtime were associated with greater social jetlag. Irregular bedtime routine, delinquent behavior, and peer delinquency were associated with later weekday bedtime. Consuming breakfast and joint (parent and child) and independent (child) decisions surrounding selecting shows and movies were associated with earlier weekday bedtimes. Delinquent behavior, peer delinquency, and difficulty falling asleep were associated with later weekend bedtimes, while joint (parent and child) and independent (child) decisions surrounding selecting shows and movies were associated with earlier weekend bedtimes. Consuming breakfast more days was associated with longer weekday time in bed, while irregular bedtime routine, peer delinquency, parent verbal aggression, difficulty falling asleep, and smartphone use near bedtime were associated with shorter weekday time in bed. Delinquent behavior was associated with shorter weekend time in bed. Conclusion Across both approaches, the most consistent findings draw attention to delinquent behavior, bedtime routine regularity, smartphone use, and difficulty falling asleep as targets to optimize adolescent sleep health. Findings have implications for clinical and policy recommendations. Support (if any)
neurosciences,clinical neurology