Preliminary appraisal of machine learning–based prediction models

Abstract

A significant portion of healthcare research is devoted to the development of prediction models, yet the integration of these models into routine clinical care remains limited. Persistent barriers, such as insufficient reporting, limited generalizability beyond the development population, and the absence of accessible, user-friendly software, continue to hinder implementation. These challenges are especially pronounced for models labeled as machine learning or artificial intelligence, which have seen a dramatic rise in popularity over the past decade. To support the responsible adoption of machine learning–based prediction models in clinical practice, this paper proposes five questions to aid in the preliminary appraisal of machine learning models published in, or submitted to, medical journals.