Application of a Bayesian strategy to ABCD: Identification of substance use risk and COVID-19 effects on neurodevelopment

Abstract Substance use initiation at an early age is associated with numerous negative outcomes, including increased likelihood of substance use disorders later in life. Multiple lines of evidence indicate that the risk for early substance use initiation is influenced by individual differences in neural development. The precise neural developmental mechanisms that give rise to heighted substance-use vulnerability remain poorly understood and The Adolescent Brain and Cognitive Development (ABCD) study provides an unprecedented opportunity to elucidate these mechanisms. However, children in this cohort now face a unique developmental challenge: entering adolescence during the COVID-19 pandemic. In direct response to PAR-19-162 ('Accelerating the Pace of Child Health Research Using Existing Data from the ABCD Study'), this application aims to characterize neurodevelopmental trajectories of substance use risk with specific consideration of the societal and individual effects of COVID-19. Specifically, using Bayesian machine learning and hierarchical time-series modeling of longitudinal ABCD data, this proposal will establish, refine, and deploy models of normative trajectories in brain development and quantify deviations related to substance-use risk (AIM 1). Further, this effort will carefully contextualize the effects of the COVID-19 crisis as a US-wide event with deep consequences for child development (AIM 2). As a primary research product of this proposal, all derived models and functional connectivity metrics will be shared via ABCD's central repository (AIM 3). This will include (i) complete neural 'fingerprints' or functional connectivity matrices for all task-based data from ages 10-14; (ii) derived normative 'growth curves', and (iii) the full generative probabilistic models for reuse by other laboratories. This key data contribution will relieve logistic burdens for a large number of research labs and further promote widespread use of ABCD data, propelling comparability and reproducibility of single-subject prediction studies towards identifying a reliable predictor of substance-use initiation in youth. This is a critical step toward precision psychiatry and will shed light on individual difference factors that contribute to vulnerability in the exigent context of the evolving COVID-19 pandemic. Such predictors are needed to understand the developmental trajectories of substance- use phenotypes and to inform early risk models and preventative intervention efforts.