PROJECT SUMMARY / ABSTRACT Nearly 30 million individuals in the US will suffer from an eating disorder in their lifetime, with onset most commonly in adolescence and with $65 billion in yearly economic costs. The most prevalent of all eating disorder phenotypes is binge-eating disorder, which affects 3-5% of the US population and portends an array of medical and psychiatric sequelae, including cardiometabolic disease and elevated suicidality. Social media use is ubiquitous among adolescents and implicated in binge-eating disorder, but the directionality and mechanisms remain unclear. Possible pathways include depression, anxiety, cyberbullying, stress, and poor sleep. Prior studies have been mostly cross-sectional, and patterns and associations may differ for diverse subpopulations (e.g., race/ethnicity, gender, sexual orientation). The prevention of binge-eating disorder requires accurate prediction. Mobile phone use patterns could predict binge-eating disorder; binge-scrolling social media could predict binge eating. Current guidance for adolescent social media use is limited and non- specific. Our long-term goal is to inform precision guidance for social media use for adolescents, parents, and clinicians to mitigate adverse mental health risks and optimize wellbeing. Our objective is to identify prospective associations, sensitive periods, and mechanisms between social media and binge-eating disorder from early to late adolescence. Our central hypothesis is that problematic social media behavior patterns (e.g., addiction, conflict, overuse, tolerance, and relapse) can predict binge-eating disorder through depression, anxiety, cyberbullying, stress, and poor sleep. To achieve our objective, we will leverage comprehensive assessments of social media use (problematic social media use and a novel objective mobile phone tracking app) and binge-eating behaviors and disorder among a diverse national prospective cohort in the Adolescent Brain Cognitive Development (ABCD) Study (N=11,875), followed annually (6 years completed). The cohort uniquely starts prior to adolescence to capture onset of social media and binge-eating patterns through early, mid, and late-adolescence. We will use robust longitudinal and machine learning methods to analyze all available years of data in the ABCD Study. Our specific aims will: 1) determine longitudinal associations, bidirectional relationships, and sensitive windows linking problematic social media use patterns and binge- eating disorder, applying an intersectional framework to characterize heterogeneity by race/ethnicity, gender, and sexual orientation; 2) identify mechanisms linking problematic social media use with binge-eating disorder; 3) use machine learning algorithms applied to mobile phone data to determine the extent to which social media use patterns over time can predict binge-eating disorder; and 4) translate findings to inform guidance for adolescents, parents, and clinicians. We will partner with a ...