More recently, polygenic scores (PGS), which aggregate risk for a trait across the genome using information from genome-wide association studies (GWAS), were robustly associated with substance use [12] and substance-related problems [13] across adolescence and into young adulthood. However, though robustly associated, current PGS do poorly in identifying individuals affected by SUDs [14]. To date, there is limited work on the combined impact of genetic, environmental, and clinical risk factors for SUDs. Prior work combining individual genetic variants and clinical features outperformed clinical features alone [15], but individual variants have limited predictive power. In other medical conditions, such as melanoma [16] or ischemic stroke [17], combining clinical and genetic risk factors showed improvement predicting risk for a specific outcome over models using individual risk factors.
