Impact of Antipsychotic Treatment on Cognition and Symptoms in First‐Episode Schizophrenia: A Machine Learning Approach

Background This study investigates antipsychotic treatment effects on cognition and symptoms in first‐episode schizophrenia (FES) and examines the impact of duration of untreated psychosis (DUP) on these outcomes, employing machine learning for predictive modeling. Methods In a prospective 8‐week study, 106 male FES patients received cognitive assessment using the MATRICS Consensus Cognitive Battery (MCCB) and clinical evaluation via the Positive and Negative Syndrome Scale (PANSS). Machine learning models, including multilayer perceptron (MLP) and random forest (RF), were applied to baseline data to predict changes in MCCB and PANSS scores. Results Treatment yielded significant cognitive and symptomatic improvements. DUP was documented from clinical records; descriptive values are not presented in this report, and no claims are made regarding DUP effects. The MLP model accurately predicted MCCB scores ( R 2 = 1), with social cognition and working memory as key features. The RF model predicted PANSS scores effectively ( R 2 = 0.72). Conclusion Antipsychotic treatment improved cognition and symptoms. DUP was recorded but not analyzed as a primary predictor in this report. Machine learning models effectively predict individualized outcomes, highlighting their potential for clinical decision support in schizophrenia.