nomROC {nonbinROC}R Documentation

Performs ROC-type analysis for nominal scale gold standard

Description

Assess the accuracies of diagnostic tests when the gold standard is nominal.

Usage

nomROC(gldstd, test1, test2 = NULL, penalty = NULL)

Arguments

gldstd vector giving the gold standard
test1 data frame giving the set of confidence scores
test2 optional data frame giving another set of confidence scores
penalty optional square matrix for the penalty function L[i,j] in which 0 <= L[i,j] <= 1 for j > i and the remaining elements can be set at any value. By default, L[i,j] = 1 for j > i and 0 otherwise.

Value

In the paired design, `nomROC' returns the pairwise accuracies and associated standard errors between each category of the gold standard, the penalty matrix, the overall accuracies and associated standard errors of the diagnostic tests, their overall covariance and the test for comparing two diagnostic tests' accuracies.
Otherwise, `nomROC' returns the pairwise accuracies and associated standard errors between each category of the gold standard, the penalty matrix, the overall accuracy and associated standard error of the diagnostic test.

Author(s)

Paul Nguyen

References

Obuchowski, N. A. (2005) Estimating and comparing diagnostic tests' accuracy when the gold standard is not binary. Academic Radiology, 12, 1198-1204.

Examples

#A paired design example for acute abdominal pain from Obuchowski, Goske and Applegate (2001)
data(abpain)
attach(abpain)
pre = data.frame(Pre1, Pre2, Pre3)
post = data.frame(Post1, Post2, Post3)
penalty = matrix(c(0,0,0,1,0,0,1,0.5,0), nrow = 3)
nomROC(Group, pre, post, penalty)

[Package nonbinROC version 1.0.1 Index]