auc: Calculate and compare AUC estimate for clustered data

In hild0146/fastAUC: Calculate and Compare AUC estimates

Description

This function can be used to calculate and compare AUC estimates. If a single set of test measurements is provided, the corresponding AUC and variance values are computed. If two sets of test measurements are provided, a significance test for a difference in AUC estimates is also computed. Optionally, the previous can also be computed in the presence of clustered data per Obuchowski (1997). This function implicity assumes that higher values of test_1 and test_2 are correlated with the positive gold standard (status = TRUE)

Usage

auc(test_1, test_2 = NULL, status, cluster = NULL)

Arguments

test_1	the set of test measurments for which the AUC should be computed
test_2	a second set of test measurments for which the AUC is computed and compared to the first set of test measurements
status	a logical vector representing the gold standard negative/positive status of an observaiton
cluster	variable identifying clusters

Value

a list containing

auc	a vector of AUC estimates
var	the individual variance estimate or covariance matrix
test_stat	the test statistic for comparing the AUC estimates, distributed standard normal
p_value	the p-value associated with the AUC test comparison (2-sided)

References

Obuchowski N.A. (1997) Nonparametric Analysis of clustered ROC Curve Data. Biometrics, 53, 567-578.

DeLong E.R., DeLong D.M, Clarke-Pearson D.L. (1988) Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics, 44, 837-845.

Examples

library(fastAUC)

n_obs <- 10000
n_clust <- 100

cluster <- sample(x = 1:n_clust, size = n_obs, replace = T)
status <- as.logical(rbinom(n = n_obs, size = 1,  prob = 0.3))
test_1 <- ifelse(status,
                 rnorm(n = n_obs, mean = 15, sd = 20),
                 rnorm(n = n_obs,mean = 10, sd = 20))
test_2 <- ifelse(status,
                 rnorm(n = n_obs, mean = 20, sd = 20),
                 rnorm(n = n_obs,mean = 10, sd = 20))

auc(test_1 = test_1, status = status)

auc(test_1 = test_1, test_2 = test_2, status = status, cluster = cluster)

hild0146/fastAUC documentation built on May 6, 2019, 12:05 a.m.