gpqyouden provides exact, simulation-based confidence intervals for the
Youden index and its optimal cut-point, using the generalized
pivotal quantity (GPQ) method of Weerahandi (1993) as developed for this
problem by Lai, Tian and Schisterman (2012). It also includes a robustness
study under a contaminated-normal mixture model, and the full Monte Carlo
simulation machinery used to evaluate coverage.
The Youden index summarises how well a continuous biomarker discriminates between two groups (e.g. diseased / healthy), and its optimal cut-point is the threshold that achieves that separation. Standard asymptotic confidence intervals for both quantities can under-cover badly in small-to-moderate samples; the GPQ method gives exact intervals by Monte Carlo simulation instead of a large-sample approximation.
This package was developed as part of a Master's thesis in Data Science (Universidad Nacional de Educación a Distancia, 2026), which reproduces and extends the simulation study of Lai et al. (2012), evaluates robustness under model contamination, and applies the method to a real biomarker dataset (Duchenne muscular dystrophy carrier detection).
# install.packages("devtools")
# devtools::install_github("jolmedilla/gpqyouden")
install.packages("gpqyouden", repos = c('https://jolmedilla.r-universe.dev', 'https://cloud.r-project.org'))
library(gpqyouden)
set.seed(1)
y1 <- rnorm(30, mean = 2, sd = 1) # diseased group
y2 <- rnorm(30, mean = 0, sd = 1) # non-diseased group
gpq <- gpq_from_sample(y1, y2, K = 2000)
# 95% confidence interval for the Youden index
quantile(gpq$RJ, probs = c(0.025, 0.975))
# 95% confidence interval for the optimal cut-point
quantile(gpq$Rc, probs = c(0.025, 0.975))
To run a full simulation study over a grid of scenarios:
grid <- default_param_grid() # the 160-scenario design of Lai et al. (2012)
results <- run_parallel_grid(grid[1:5, ], n_mc = 200, K = 2000)
Lai, C.-Y., Tian, L., & Schisterman, E. F. (2012). Exact confidence interval estimation for the Youden index and its corresponding optimal cut-point. Computational Statistics & Data Analysis, 56(5), 1103–1114. https://doi.org/10.1016/j.csda.2010.11.023
Weerahandi, S. (1993). Generalized confidence intervals. Journal of the American Statistical Association, 88(423), 899–905. https://doi.org/10.2307/2290779
MIT © Juan José Olmedilla Arregui