Ensuring generalizability and clinical utility in mental health care applications: Robust artificial intelligence‐based treatment predictions in diverse psychosis populations

Abstract

We present a robust framework for training and assessing the clinical utility of prediction models in psychiatry. Our models generalize across different psychosis populations and show promising calibration and net benefit. However, performance disparities across demographic and treatment subgroups highlight the need for more diverse clinical samples to ensure equitable prediction.

Ähnliche Arbeiten