The objective of this study is to develop models to predict the development of substance use disorders (SUDs) among adolescents in four health systems: KPNC, KPHI, Henry Ford and Geisinger. We will focus on EHR data available during study period from birth cohorts with dates of birth occurring during 1997-2003 to examine the predictors of developing a SUD between the ages of 12-18. We will use the data infrastructure and resources of the Addiction Research Network of the HCSRNs VDW to develop predictive models for the development of adolescent SUDs. The four systems offer significant geographic, demographic, delivery design, and socioeconomic and epidemiological diversity, enhancing the generalizability of findings while also providing significant coverage of the national population. The goal is to identify high risk children for feasible, cost-effective, targeted prevention and intervention.