ABSTRACT Rates of alcohol and drug use rise sharply throughout adolescence (ages 10-21), peak in young adulthood, and then decline. Approximately 10% of adolescents in the US report consuming alcohol in the past month1. By the eighth grade, approximately 20% of youth in the US have tried cannabis and 50% of youth have done so by the 10th grade1. The highest rates of heavy alcohol and illicit substance use occur during adolescent years2. This increase in substance use and misuse underscores the importance of understanding vulnerability for substance use-related behaviors during adolescence. Substantial research links neurocognitive processes, such as response inhibition, working memory, and cost-benefit decision-making, to behaviors related to the misuse of substances (e.g., binge drinking, sharing needles) and substance use disorders. Prospective studies suggest that individual differences in neurocognition related to cognitive control and affective sensitivity, in particular, contribute to the risk for substance use behavior in adolescence. However, this work has been limited in its ability to combine multiple datatypes, such as neuroimaging metrics and behavioral data, purportedly tapping an underlying latent feature (i.e., neurocognitive capacity) and extract `fingerprints' that are useful for individual prediction. This proposal plans to use data from the Adolescent Brain Cognitive Development℠ (ABCD) Study, which collects annual measures of neurocognition for approximately 12,000 youth, to implement an innovative multilevel methodological framework in order to (1) develop a Bayesian non-parametric dimensionality reduction algorithm toolkit that allows for the reduction of a multidimensional dataset (e.g., mean reaction times, brain connectivity measures across multiple tasks) into a much smaller collection of latent features and (2) use the extracted latent features to predict the onset, maintenance, and nature of substance use in developing adolescents. By precisely identifying and specifying the variation in underlying cognitive-affective, behavioral, and neurobiological mechanisms, this proposal offers an innovative empirical foundation for the development of prevention and treatment methods that address important clinical endpoints such as substance use-related problems. Moreover, this research provides a framework for promoting the initiation, personalization, and maintenance of behavior change by integrating work across theoretical and methodological domains.