PROJECT SUMMARY There is increasing evidence that early intervention for psychosis in coordinated specialty care (CSC) services improves outcomes and lives. The outcome of early course psychosis (EP) is heterogeneous, ranging from early full recovery to treatment resistance and functional decline from the onset of illness. This heterogeneity limits our ability to predict individual level outcomes needed for treatment planning and for tailoring the type, duration and intensity of therapeutic interventions. Biomarkers as well as clinical and demographic features, early in the illness can predict outcome, but taken individually, their prognostic value is limited. Our Bipolar- Schizophrenia Network for Intermediate Phenotypes (BSNIP) consortium has recently developed, replicated and validated a biomarker (EEG, eye movement testing, and neurocognition) based categorization (Biotypes 1, 2 and 3) in a trans-diagnostic sample of cases with idiopathic psychosis (schizophrenia, schizoaffective disorder, or bipolar disorder with psychosis), ranging from 18-35 years of age. In this study, we will leverage this categorization, along with clinical and biomarker data to predict illness trajectory and outcome during follow-up at 1, 6 and 12 months in 320 EP patients across CSC clinics at the five B-SNIP sites. First, we will characterize outcome trajectories and Biotype structure in EP. Our available data indicate the Biotype structure will be the same in EP as in our large sample. Second, we will investigate the predictive value of the nine bio-factors and the three Biotypes identified by B-SNIP for symptomatic and functional outcome. We predict that the EP population will manifest distinct outcome clinical trajectories (good, intermediate and poor) and will have a Biotype structure similar to that seen in chronic psychosis subjects, i.e., Biotypes 1, 2 and 3) (hypothesis 1). Biotype-3, and Biotye-2 cases, will have the best outcomes (defined both categorically, and dimensionally, using symptomatic, cognitive and functional measures); Biotype-1 will have the worst outcomes to CSC treatment, across all target time points (hypothesis 2). Notably, Biotype-1 and Biotype-2 cases will have the same level of cognition function at baseline. Finally, we will investigate the predictive value of clinical (such as diagnosis, illness duration, substance abuse, and treatment adherence), and biomarker (including neuroimaging) features in a multi-variate model and will develop a feasible biomarker battery and predictive algorithm for application in community CSC sites nation-wide. We will thus provide to the field a means for predicting success of EP cases in CSC treatment to improve clinical practice and to enhance efficient use of available treatment resources.