Internal-External Cross-Validation (IECV) Matrix — consortium_cross_site

Evaluates model transportability across multi-hospital network studies by fitting models on Site i and evaluating on Site j for all (i, j) pairs in K x K consortium matrix.

Usage

consortium_cross_site_validation(
  site_data,
  feature_cols,
  target_col,
  site_col = "site",
  regularize = TRUE,
  alpha = 1
)

Arguments

site_data: Data frame containing predictive features, target, and site_col, or named list of data frames per site.
feature_cols: Character vector of feature column names.
target_col: Name of the binary target outcome column.
site_col: Name of the site identifier column (default "site").
regularize: Logical; whether to apply L2 ridge regularization (default TRUE).
alpha: Numeric; L2 penalty parameter (default 1.0).

Value

An S3 object of class consortium_cross_site_validation containing $auroc_matrix, $brier_matrix, and $summary.