Plot Time-Dependent ROC Curves

Generates time-dependent ROC curves for evaluating prognostic accuracy of survival models. Plots training and testing ROC curves at the 90th percentile survival time.

Usage

PlotTimeROC(
  train.x,
  train.y,
  test.x,
  test.y,
  model,
  modelname,
  cols = NULL,
  palette = "jama"
)

Arguments

train.x

Matrix or data frame of training predictors.

train.y

Training survival outcomes (time and status).

test.x

Matrix or data frame of testing predictors.

test.y

Testing survival outcomes (time and status).

model

Fitted survival model object.

modelname

Character string for model identification.

cols

Optional vector of colors for plotting.

palette

Character string specifying color palette. Default is `"jama"`.

Value

A `ggplot` object representing the ROC curve plot.

Author

Dongqiang Zeng

Examples

if (requireNamespace("glmnet", quietly = TRUE) &&
  requireNamespace("survival", quietly = TRUE) &&
  requireNamespace("timeROC", quietly = TRUE)) {
  library(survival)
  set.seed(123)
  train_x <- matrix(rnorm(100 * 5), ncol = 5)
  train_y <- data.frame(time = rexp(100), status = rbinom(100, 1, 0.5))
  test_x <- matrix(rnorm(50 * 5), ncol = 5)
  test_y <- data.frame(time = rexp(50), status = rbinom(50, 1, 0.5))
  fit <- glmnet::cv.glmnet(train_x, Surv(train_y$time, train_y$status), family = "cox")
  p <- PlotTimeROC(train_x, train_y, test_x, test_y, fit, "Cox Model")
  print(p)
}