Official implementation of the paper:
Integrating Epigenetic and Phenotypic Features for Biological Age Estimation in Cancer Patients via Multimodal Learning
Shuyue Jiang, Wenjing Ma, Shaojun Yu, Chang Su, Runze Yan, Jiaying Lu
arXiv preprint, 2025

This project is designed to run on Google Colab with a T4 GPU.
Additional dependencies can be installed as shown in the provided code scripts.

Our pipeline includes a feature selection step using BorutaShap.
Due to environment limitations in Google Colab, this step should be performed locally.

The framework adopts a modular architecture, allowing flexible model choices. We implement two variants: one using the lightweight ElasticNet and the other leveraging the large-scale foundation model TabPFN.

Example scripts for running EPICAGE-TabPFN are provided in EPICAGE. ipynb, and scripts for EPICAGE-ElasticNet are provided in EPICAGE-ElasticNet.ipynb.

Additional details are provided in the Appendix.