tdrocc: Function to compute time-dependent ROC curves

Description

The function is a wrapper for the survivalROC function in order to compute the time-dependent ROC curves.

Usage

tdrocc(x, surv.time, surv.event, surv.entry = NULL, time, cutpts = NA, na.rm = FALSE, verbose = FALSE, span = 0, lambda = 0, ...)

Arguments

vector of risk scores.

surv.time

vector of times to event occurrence.

surv.event

vector of event occurrence indicators.

surv.entry

entry time for the subjects.

time

time point for the ROC curve.

cutpts

cut points for the risk score.

na.rm

TRUE if the missing values should be removed from the data, FALSE otherwise.

verbose

verbosity of the function.

span

Span for the NNE, need either lambda or span for NNE.

lambda

smoothing parameter for NNE.

...

additional arguments to be passed to the survivalROC function.

Value

spec: specificity estimates
sens: sensitivity estimates
rule: rule to compute the predictions at each cutoff
cuts: cutoffs
time: time point at which the time-dependent ROC is computed
survival: overall survival at the time point
AUC: Area Under the Curve (AUC) of teh time-dependent ROC curve
data: survival data and risk score used to compute the time-dependent ROC curve

References

Heagerty, P. J. and Lumley, T. L. and Pepe, M. S. (2000) "Time-Dependent ROC Curves for Censored Survival Data and a Diagnostic Marker", Biometrics, 56, pages 337--344.

Examples

Run this code

set.seed(12345)
age <- rnorm(100, 50, 10)
stime <- rexp(100)
cens <- runif(100,.5,2)
sevent <- as.numeric(stime <= cens)
stime <- pmin(stime, cens)
tdroc <- tdrocc(x=age, surv.time=stime, surv.event=sevent, time=1,
  na.rm=TRUE, verbose=FALSE)
##plot the time-dependent ROC curve
plot(x=1-tdroc$spec, y=tdroc$sens, type="l", xlab="1 - specificity",
  ylab="sensitivity", xlim=c(0, 1), ylim=c(0, 1))
lines(x=c(0,1), y=c(0,1), lty=3, col="red")