Machine learning applications hold promise for creating innovative disease prediction models based on longitudinal data. This study used COGA’s rich datasets with EEG, genetic, and FH information acquired from individuals as early as age 12, before developing AUD, and followed years later when they either were diagnosed as DSM-5 AUD or unaffected. This is the first study to formulate a prediction model for those who are predisposed to develop AUD using ML with multidimensional features while considering gender and ancestry. We found higher accuracy rates for the prediction models in AA than EA samples. In both AA and EA samples combining EEG and SNP features resulted in higher accuracy scores than the models based on only EEG features or only SNPs, and these results were confirmed in a follow up analysis (same dataset) within the different AA age groups (early adolescence, late adolescence and adults) and EA late adolescence age group. Gender analyses revealed trend of higher model accuracy in the female group over the male group in both the EA and the AA for all three features categories (EEG,
