Mass Multivariate Derivation and Validation of AUD Biotypes using Developmental Imaging and Genomic Approaches

PROJECT SUMMARY/ABSTRACT In stark contrast to the widespread prevalence and devastating outcomes associated with alcohol use disorder, currently available treatment options are only moderately effective. The large heterogeneity in AUD presentations may obfuscate etiology and individualized treatment options. In this 5-year K01 mentored research scientist award application, I propose to characterize the behavioral, molecular, and genetic correlates of AUD biotypes (subtypes determined by biology) across the lifespan. To this end, I will apply semi-supervised machine learning algorithms to structural brain imaging data from the largest available AUD and alcohol use datasets from childhood to old age (total n=61,428). I will examine the stability of these biotypes across the lifespan, including among substance-naïve children and adults with heavy alcohol use, as well as their correlates with neurobehavioral stage-based constructs of addiction (i.e., impulsivity, negative affect, cognition) and with alcohol involvement trajectories. I will then conduct a genome-wide association study of AUD biotypes to disarticulate their genetic architecture, genetic correlates, and potential molecular pathways that may be leveraged for drug repositioning. This grant develops my skillsets in semi-supervised machine learning, AUD heterogeneous presentations, multivariate genome-wide methods, and multi-omic analytic approaches. These skill sets will serve as a backdrop for a planned R01 submission that will leverage my background in large-scale data analysis to translate across biological modalities in substance use research.