0. One-Click Run All Models

In E2E: Ensemble Learning Framework for Diagnostic and Prognostic Modeling

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.width = 10,
  fig.height = 8
)

library(E2E)

# Set up parallel processing
cl <- parallel::makeCluster(2)
doParallel::registerDoParallel(cl)

Overview

E2E provides three powerful one-click functions that automatically run comprehensive modeling pipelines:

• int_dia(): Diagnostic modeling pipeline (52 model variants)
• int_imbalance(): Imbalanced data diagnostic pipeline (64 model variants)
• int_pro(): Prognostic modeling pipeline (24 model variants)

---

1. Diagnostic Pipeline

# Run all diagnostic models
results_dia <- int_dia(
  train_dia,
  test_dia,
  test_dia, #can be any other data
  tune = FALSE,
  n_estimators = 5,
  seed = 123
)

# Visualize results
#plot_integrated_results(results_dia, metric_name = "AUROC")

---

2. Imbalanced Data Pipeline

# Run all models including imbalance handling methods
results_imb <- int_imbalance(
  train_dia,
  test_dia,
  test_dia, #can be any other data
  tune = FALSE,
  n_estimators = 5,
  seed = 123
)

# Visualize results
#plot_integrated_results(results_imb, metric_name = "AUROC")

---

3. Prognostic Pipeline

# Run all prognostic models
results_pro <- int_pro(
  train_pro,
  test_pro,
  test_pro, #can be any other data
  tune = FALSE,
  n_estimators = 5,
  time_unit = "day",
  years_to_evaluate = c(1, 3, 5),
  seed = 123
)

# Visualize results (C-index)
#plot_integrated_results(results_pro, metric_name = "C-index")

---

Key Features

✅ Fully Automated: Run dozens of models with one line of code ✅ Multi-Dataset Evaluation: Evaluate on training and multiple test sets simultaneously ✅ Diverse Methods: Covers single models, Bagging, Stacking, and Voting ✅ Clear Visualization: Heatmaps display all results intuitively

# Stop parallel cluster
parallel::stopCluster(cl)

E2E documentation built on March 22, 2026, 1:07 a.m.