Researchers developed a clinical prediction model using electronic health records from over 127,000 patients to identify individuals at risk of psychotic or bipolar disorders. The model demonstrated high discriminative power (C-index 0.80) and practical utility for early intervention strategies.