loki2.mil.models.TCGA-Loki2_MILCell_blca_metastasis_5fold

5-fold cross-validation script for BLCA metastasis MIL model.

This script performs 5-fold cross-validation training for a Multiple Instance Learning (MIL) model to predict metastasis status from cell embeddings.

Module Contents

loki2.mil.models.TCGA-Loki2_MILCell_blca_metastasis_5fold.main(args: argparse.Namespace) → None

Main function for 5-fold cross-validation training.

Performs data preparation, creates balanced folds, trains models for each fold, and computes overall cross-validation metrics.

Parameters:: args – Argument namespace containing all hyperparameters and paths: - data_path: Path to embedding data parquet file. - type_path: Path to label/metadata file. - base_output_folder: Base output folder for results. - class_type: Class type name (e.g., “Disease_Stage”). - signature: Signature name (e.g., “Patient_Label”). - n_folds: Number of folds for cross-validation. - seed: Random seed. - All other model and training hyperparameters.