EP01.05-008 Pre-Trial Validation of a Risk Prediction Model for Pulmonary Nodules Trained on a Nested Case-Control Study with No External Cohort

Abstract

Prior to testing in a clinical trial, a biomarker for diagnostic risk prediction requires validation through a nested case-control study design to avoid verification and other biases and split or independent samples to avoid overfitting bias. When the focus within this paradigm is on building a risk prediction model, standard metrics include discrimination, measured by the area under the ROC curve (AUC) and calibration measured by the accuracy of absolute risk estimates using calibration curves, intercept, and slope.