roc.area.test: Nonparametric area under the ROC curve
Description
Computes the nonparametric area under the ROC curve and its variance
based on U-statistic theory (DDCP).
Usage
roc.area.test(markers, status)
# S3 method for roc.area.test
print(x, ...)
Value
a list with the following elements
area
estimated area.
var
estimated variance (matrix).
stat
test statistic for equality of AUCs. Is not returned when
only one diagnostic marker is present.
p.value
the p-value for the test of equality (2-sided).
df
the degrees of freedom of the chi-square.
The "print" method formats and returns the output.
Arguments
markers
The marker values for each subject. If there are more
than one markers then this should be a matrix.
status
binary disease status indicator
x
object of class roc.area.test output from this function.
...
optional arguments to the print function.
Details
It calculates the area and its variance. For more than one marker it
calculates the statistic to test for the equality of all AUCs. This
statistic has a standard normal reference distribution for two
variables and chi-square with number of variables minus 1.
References
DeLong, E. R., D. M. DeLong, and D. L. Clarke-Pearson. 1988. Comparing
the areas under two or more correlated receiver operating characteristic
curves: A nonparametric approach. Biometrics 44:837-845.
g <- rep(0:1, 50)
x <- rnorm(100) + g
y <- rnorm(100) + g
z <- rnorm(100) + g
roc.area.test(cbind(x,y), g)
roc.area.test(cbind(x,y,z), g)
y1 <- y + 0.75*g
roc.area.test(cbind(x,y1), g)