---
tags:
- Survival-Analysis
---
<p align="center">
  <h2 align="center">[Under review] Replicating Patient Follow-Up with Hierarchical Directed Graphs for Head and Neck Cancer Survival Analysis 🧪🔬🎯</h2>
</p>

✅ Official HuggingFace repository of the paper "Replicating Patient Follow-Up with Hierarchical Directed Graphs for Head and Neck Cancer Survival Analysis".

📄 Preprint, under review for MIDL 2026: [arXiv preprint coming soon].

### 🧩 Method Overview

We propose **H2DGSurv** (Hierarchical Directed Heterogeneous Graph), a Graph Neural Network architecture for multimodal survival prediction that models the clinical pathway as a directed heterogeneous graph with temporal progression.

<p align="center">
  <img src="./figures/method_overview.png" alt="Method Overview" width="800">
</p>

### 📂 Available resources

**h2dg.pt**: PyTorch model weights.

**folds_5.csv**: Pandas DataFrame indicating cross-validation splits for each of the 5 folds.

### 🚀 Source code

PyTorch model implementation is available on github: [source code](https://github.com/dpmc-lab/h2dg-surv)

### 🙌 Acknowledgments

We acknowledge [Kist et al. 2024](https://www.nature.com/articles/s41597-024-03596-3) for making the HANCOCK dataset available.

### Useful Links

- [HANCOCK Challenge](https://www.hancock.research.uni-erlangen.org/download)
- [BioClinicalBERT](https://huggingface.co/emilyalsentzer/Bio_ClinicalBERT)

---

### 🔗 Citation

> [!IMPORTANT]  
> This project is based on the work by Miccinilli and Di Piazza 2025. If you use this code in your research, we would appreciate reference to the following paper:

```BibTeX
@inproceedings{mcdp2025h2dg,
  author    = {Hugo Miccinilli and Theo Di Piazza},
  title     = {Replicating Patient Follow-Up with Hierarchical Directed Graphs for Head and Neck Cancer Survival Analysis},
  booktitle = {Arxiv preprint},
  year      = {2025},
}
```