Skip to contents

Stability Selection for KL-Integrated Cox Elastic-Net Models

coxkl_enet.StabSelect.Rd

Performs stability selection for the KL-integrated Cox elastic-net model by repeatedly refitting the model on bootstrap or subsampled datasets and aggregating variable selection frequencies across replicates. This procedure provides a robust measure of variable importance that is less sensitive to a single split of the data.

Usage

coxkl_enet.StabSelect(
  z,
  delta,
  time,
  stratum = NULL,
  RS = NULL,
  beta = NULL,
  etas,
  alpha = 1,
  lambda = NULL,
  nlambda = 100,
  lambda.min.ratio = 0.1,
  nfolds = 5,
  cv.criteria = c("V&VH", "LinPred", "CIndex_pooled", "CIndex_foldaverage"),
  c_index_stratum = NULL,
  message = FALSE,
  seed = NULL,
  B = 50,
  fraction_sample = 0.5,
  ncores = 1,
  ...
)

Arguments

z

Numeric matrix of covariates. Rows represent individuals and columns represent predictors.

delta

Numeric vector of event indicators (1 = event, 0 = censored).

time

Numeric vector of observed times (event or censoring).

stratum

Optional numeric or factor vector indicating strata. If NULL, all subjects are assumed to be in the same stratum.

RS

Optional numeric vector or matrix of external risk scores. If not provided, beta must be supplied.

beta

Optional numeric vector of external coefficients (length equal to ncol(z)). If provided, it is used to compute external risk scores. If not provided, RS must be supplied.

etas

Numeric vector of candidate eta values to be evaluated.

alpha

Elastic-net mixing parameter in \((0,1]\). Default is 1 (lasso penalty).

lambda

Optional numeric vector of lambda values. If NULL, a path is generated automatically.

nlambda

Integer. Number of lambda values to generate if lambda is NULL. Default is 100.

lambda.min.ratio

Numeric. Ratio of min/max lambda. Default depends on sample size vs dimension (0.05 if n < p, else 1e-03).

nfolds

Integer. Number of cross-validation folds. Default is 5.

cv.criteria

Character string specifying the cross-validation criterion for selecting both eta and lambda. Choices are:

  • "V&VH" (default): V&VH loss.

  • "LinPred": Loss based on cross-validated linear predictors.

  • "CIndex_pooled": Pooled C-Index.

  • "CIndex_foldaverage": Average C-Index across folds.

c_index_stratum

Optional stratum vector. Required only when cv.criteria is set to "CIndex_pooled" or "CIndex_foldaverage", and a stratified C-index needs to be computed while the fitted model is non-stratified. Default is NULL.

message

Logical. Whether to print progress messages. Default is FALSE.

seed

Optional integer. Random seed for reproducible fold assignment.

B

Integer. Number of bootstrap/subsampling replicates used for stability selection. Default is 50.

fraction_sample

Numeric in (0, 1]. Fraction of the original sample size used in each replicate (without replacement if subsampling is used). Default is 0.5.

ncores

Integer. Number of parallel cores. Default 1 (sequential execution).

...

Additional arguments passed to the underlying fitting function coxkl_enet.

Value

An object of class "StabSelect", which is a list containing:

stability_path

A numeric matrix of dimension n_vars x n_lambda giving, for each variable (rows) and each value of lambda (columns), the empirical selection probability across the B replicates.

lambda

Numeric vector giving the global lambda sequence used for the underlying elastic-net fits.

Examples

if (FALSE) { # \dontrun{
data(ExampleData_highdim)
train_dat_highdim      <- ExampleData_highdim$train
beta_external_highdim  <- ExampleData_highdim$beta_external

eta_list <- generate_eta(method = "exponential", n = 10, max_eta = 50)

coxkl.StabSelect <- coxkl_enet.StabSelect(
  z            = train_dat_highdim$z,
  delta        = train_dat_highdim$status,
  time         = train_dat_highdim$time,
  stratum      = train_dat_highdim$stratum,
  beta         = beta_external_highdim,
  etas         = eta_list,
  cv.criteria  = "CIndex_pooled",
  B            = 20,
  message      = TRUE
)

# Plot with different thresholds without re-running stability selection
plot(coxkl.StabSelect, threshold = 0.6)
plot(coxkl.StabSelect, threshold = 0.8)
} # }