Multimodal Late Fusion with 'caret'

Extends the 'caret' framework to support late fusion workflows, enabling users to train models independently across multiple data modalities and combine their predictions into a single meta-model. Designed for developers, data scientists, and biomedical researchers alike, 'caretMultimodal' aims to make late fusion ensemble modelling as accessible and flexible as single-dataset workflows in 'caret'. Late fusion methods are based on Wolpert (1992) .

caretMultimodal

caretMultimodal extends the caret framework to support late fusion workflows in R, enabling users to train models independently across multiple data modalities and combine their predictions into a single meta-model. Designed for R developers, data scientists, and biomedical researchers, caretMultimodal makes late fusion ensemble modelling as accessible and flexible as single-dataset workflows in caret.

Example late fusion workflow using cross-validation

Key Features

• Includes all the functionality of caret, giving users full control over sampling strategies, training methods, hyperparameter tuning, and more
• Default cross-validation structure with careful handling to prevent data leakage across modalities
• Late fusion ensembling using stacked generalization
• Parallelization for faster training across models and datasets
• Model trimming to reduce memory usage for large ensembles
• Built-in evaluation tools for performance assessment, ROC curves, and variable importance
• Detailed error messages to simplify debugging

Documentation

Full API documentation is available at compbio-lab.github.io/caretMultimodal

Installation

The package can be installed using devtools

devtools::install_github("CompBio-Lab/caretMultimodal")

Acknowledgements

The project structure is inspired by Zach Mayer's caretEnsemble package, which is used for stacking multiple models on a single dataset.