ROC-Based Inference for Diagnostic Accuracy Under Verification Bias

Provides point estimates and confidence intervals for receiver operating characteristic (ROC)–based diagnostic accuracy metrics for tests and biomarkers subject to verification bias. Supported metrics include the Area Under the ROC Curve (AUC), the Youden index, and the sensitivity at a user‑specified specificity level for two‑class continuous tests under missing‑at‑random (MAR) disease verification. Point estimation follows Alonzo and Pepe (2005) . Multiple types of confidence intervals are implemented and compared, including bootstrap‑based, Method of Variance Estimates Recovery (MOVER)–based, and empirical likelihood (EL)–based intervals; see Wang et al. (2025) and < https://github.com/swang1021/rocvb>.

rocvb: R package for ROC inference under verification bias

Overview

rocvb provides point estimates and confidence intervals for ROC‑based diagnostic accuracy metrics for tests and biomarkers subject to verification bias.

rocvb currently supports inference for the following accuracy metrics for two-class continuous tests under missing-at-random (MAR) disease verification:

• Area under the ROC curve (AUC)
• Sensitivity at a fixed level of specificity
• Maximum Youden index

Multiple types of confidence intervals are provided, including bootstrap-based, MOVER-based, and empirical likelihood–based intervals.

For each estimand, results are returned simultaneously using four bias-corrected estimators for sensitivity and specificity proposed by Alonzo and Pepe (2005):

• FI: Full imputation
• MSI: Mean score imputation
• IPW: Inverse probability weighting
• SPE: Semiparametric efficient estimation

Installation

Install from GitHub:

install.packages("remotes")
remotes::install_github("swang1021/rocvb")

Technical details

See the function documentation and help in R for statistical details.

Usage

library(rocvb)

set.seed(123)
n <- 100
T <- abs(rnorm(n))
A <- abs(rnorm(n))

score <- 0.3 * T + 0.3 * A + rnorm(n, sd = 1)
D <- as.logical(score > stats::quantile(score, 0.7))
D[sample(n, 30)] <- NA

auc.ci.mar(T, D, A, n.boot = 50, plot = FALSE)
sen.ci.mar(T, D, A, p = 0.8, n.boot = 50, plot = FALSE)
yi.ci.mar(T, D, A, n.boot = 50, plot = FALSE)

#For more details, see
?auc.ci.mar
?sen.ci.mar
?yi.ci.mar

References

Alonzo, T. A. and Pepe, M. S. (2005). Assessing accuracy of a continuous screening test in the presence of verification bias. Journal of the Royal Statistical Society: Series C (Applied Statistics).

Wang, S., Shi, S., and Qin, G. (2025). Interval estimation for the Youden index of a continuous diagnostic test with verification biased data. Statistical Methods in Medical Research.

Wang, S., Shi, S., and Qin, G. (2025). Empirical likelihood inference for the area under the ROC curve with verification-biased data. Manuscript under peer review.

Wang, S., Shi, S., and Qin, G. (2025). Empirical likelihood-based confidence intervals for sensitivity of a continuous test at a fixed level of specificity with verification bias. Manuscript under peer review.