Principles of Clinical Prediction Model Development and Validation

Clinical prediction models help inform health care professionals, patients, and their families about the risks of disease or a future outcome and guide decision-making aimed at mitigating these risks. In this article, we review the principles of clinical prediction model development and validation, including predictor selection, model performance measures, internal validation, translation of a model to a scoring rule for clinical application, and external validation. We illustrate each principle using a published example of a prediction model developed to predict the presence of invasive bacterial infection in well-appearing febrile infants aged no more than 60 days presenting to emergency departments.