Clinical Prediction Models: Foundational Concepts

Background: Clinical prediction requires formalizing uncertainty into a statistical model. However, persistent confusion between prediction and inference, and between traditional (stepwise) and modern (penalized) development strategies, leads to unstable, poorly calibrated, and overfit models. A structured statistical framework is essential. Methods: This article is a structured, didactic tutorial that explains the core concepts of clinical prediction. It covers the definition of a prediction model, the fundamental strategies for its construction, and the essential framework for its evaluation. Results: The tutorial demystifies model construction by contrasting robust modern methods (penalized regression, LASSO) against traditional approaches (univariable filtering, stepwise selection). It explains how to manage key pitfalls such as collinearity (VIF), non-linearity (RCS), and interaction terms. Finally, it provides a comprehensive assessment framework by reframing model performance into its three essential domains: discrimination (ranking ability), calibration (probabilistic honesty), and validation (generalizability). Conclusions: This guide provides clinicians with the essential methodological foundation to critically appraise and understand modern prediction models.