Fig. 4: Baseline prediction of remission outcomes at 1-year (A), 5-year (B), and at any point across all follow-up periods (C).

Cross-validation ratio (CV-Ratio) of feature weights indicating most consistently selected variables is presented (x-axis) and colored bars indicate those that are significant using the sign-based consistency measure. A At 1-year (data available from Santander and Sao Paulo; Table S1), the SG2 membership positively predicted remission in combination with increased diagnoses of schizophreniform disorder, increased single marital status, and less schizophrenia diagnoses and unemployment. B At 5-years (London/Sao Paulo; Table S1), SG2 significantly predicted remission together with a psychosis “not otherwise specified” diagnosis, female sex, less schizophrenia diagnoses, and less single marital status. C Prediction of remission included increased single marital status, schizophreniform or brief psychotic disorder diagnoses, female sex, and global assessment of functioning (GAF) in addition to less schizophrenia diagnoses, negative symptoms, and major depressive disorder. Increased SG2 was associated with remission, while decreased SG1 membership (relative to the other variables in a multivariate analysis), was associated with remission. DUP duration of untreated psychosis, PANSS positive and negative syndrome scale, GAF global assessment of functioning, CPZ chlorpromazine equivalent dose, MDD major depressive disorder, Psychosis NOS Psychosis not otherwise specified.