Using AI to unpack the bi-directional association between adolescent social media exposure and mental health and well being - ABSTRACT With notable increases in rates of adolescent depression over the last decade, many researchers are looking to understand the causes, with social media being a main focus. Although the relation is far from conclusive, a sufficient number of studies have emerged to suggest harmful mental health impacts that further inquiry of adolescent social media use seems warranted. Current findings are conflicting, perhaps in part because the relation is bi-directional. Published research does not well speak to this hypothesis. In response to RFA-MH-25-205, “Bidirectional Influences Between Adolescent Social Media Use and Mental Health,” we propose to recruit and survey nationally 2,000 13-16 year-olds and follow them prospectively as they age into middle and later adolescence, when they are 16-19 years of age. Youth will complete eight comprehensive online surveys: One at baseline and another every 3-months thereafter for the first year, and every 6 months for the subsequent 1.5 years. Brief snapshots of social media and mood will be collected quarterly during 4-day Ecological Momentary Assessments (EMAs). Objective measures of adolescent Instagram use, the 2nd most commonly used social media app among youth people, will be continuously collected via Instagram’s API and coded using AI. Our research team will achieve the following: Specific Aim 1: Conduct focus groups to hear from youth the experiences and features of both their online and in-person worlds that they think are impacting their mental health. To ensure that online survey instrument queries all relevant exposures and outcomes, we will conduct 6 national, online focus groups with 120 youth. Their feedback will be integrated into the final online survey instrument. Specific Aim 2. Recruit and survey 2,000 13-16 year-olds living in the United States. Our main exposure of interest will be objective measures of Instagram use. Secondary exposure measures will be social media use as measured in the EMAs and online surveys. Our main outcomes of interest will be scale-based measures of: 1) Depression, 2) suicidal ideation and 3) suicidal behavior assessed in the online surveys. Secondary outcome measures will include single-item EMA-collected: 1) happiness and 2) life satisfaction. Specific Aim 3. Use longitudinal data to investigate the bidirectional impact that one’s mental health may have on youth social media use, and how their social media use may impact youth mental health. Guided by the Socio-Ecological Model, we will examine how potentially influential factors across the social ecology, particularly those that are in-person, contextualize observed relations between social media use and mental illness and health over time. Study findings will inform prevention efforts by guiding content that helps youth achieve healthy social media habits, as well as mental health promotion efforts that likely need to take a holistic view of youth and provide prevention efforts aimed at a young person’s whole experience, not just their online – or offline – world.
