The package helps to compute kernel-smoothed receiver operating characteristic (ROC) curves and area under the curve (AUC), including confidence intervals, using several bandwidth selection methods. Also provides the Youden index with confidence intervals, the Youden point, and optimal diagnostic cutoff estimation. Supports Gaussian, Biweight, and Epanechnikov kernels.

The receiver operating characteristic (ROC) curve is one of the most widely used tools for evaluating diagnostic and prognostic biomarkers across diverse scientific fields, particularly in medicine. Despite its ubiquity, ROC estimation and testing methods differ substantially in their assumptions and resulting curve properties. This package provides a unified framework for constructing, visualizing, and comparing parametric, nonparametric, semiparametric, and Bayesian ROC curves. 'ROCModels' helps researchers identify and implement ROC inference methods most suitable for their data.