cv_survdnn: K-Fold Cross-Validation for survdnn Models

In survdnn: Deep Neural Networks for Survival Analysis with R 'torch'

K-Fold Cross-Validation for survdnn Models

Description

Performs cross-validation for a 'survdnn' model using the specified evaluation metrics.

Usage

cv_survdnn(
  formula,
  data,
  times,
  metrics = c("cindex", "ibs"),
  folds = 5,
  .seed = NULL,
  .device = c("auto", "cpu", "cuda"),
  .threads = NULL,
  na_action = c("omit", "fail"),
  verbose = TRUE,
  ...
)

Arguments

formula	A survival formula, e.g., 'Surv(time, status) ~ x1 + x2'.
data	A data frame.
times	A numeric vector of evaluation time points.
metrics	A character vector: any of '"cindex"', '"brier"', '"ibs"'.
folds	Integer. Number of folds to use.
.seed	Optional. Set random seed for reproducibility.
.device	Character string indicating the computation device used when fitting the models in each fold. One of '"auto"', '"cpu"', or '"cuda"'. '"auto"' uses CUDA if available, otherwise falls back to CPU.
.threads	Optional positive integer. If provided, sets Torch CPU thread count before each fold fit via 'torch::torch_set_num_threads()'.
na_action	Character. How to handle missing values within each fold: '"omit"' drops incomplete rows; '"fail"' errors if any NA is present.
verbose	Logical; whether to print cross-validation progress and propagate verbose messages to fitting/evaluation in each fold (default: TRUE).
...	Additional arguments passed to [survdnn()].

Value

A tibble containing metric values per fold and (optionally) per time point.

Examples

if (requireNamespace("torch", quietly = TRUE) && torch::torch_is_installed()) {
  veteran <- survival::veteran
  cv_survdnn(
    survival::Surv(time, status) ~ age + karno + celltype,
    data = veteran,
    times = c(30, 90, 180),
    metrics = "ibs",
    folds = 3,
    .seed = 42,
    hidden = c(16, 8),
    epochs = 5
  )
}

survdnn documentation built on April 29, 2026, 9:07 a.m.