Multi-Channel Hypergraph Network for Sequential Diagnosis Prediction in Healthcare

This is the code repository for the paper "Multi-Channel Hypergraph Network for Sequential Diagnosis Prediction in Healthcare".

Requirements

python==3.6.13 torch==1.10.1 tqdm==4.62.3 scipy==1.5.4 scikit-learn==1.3.1 numpy==1.24.3 torch-geometric==2.2.0 torch-scatter==2.0.9 torch-sparse==0.6.12

1. Prepare the datasets

(1) Download the MIMIC-III and MIMIC-IV datasets

The original datasets can be downloaded from the following URLs.

Go to https://mimic.physionet.org/ for access. Once you have the authority for the dataset, download the dataset and extract the csv files to data/mimic3/raw/ and data/mimic4/raw/ in this project.

After the datasets are downloaded, please put each of them into a specified directory of the project.

Folder Specification

• data/
  • mimic3
    • encoded
    • parsed
    • raw
    • standard
  • mimic4
    • encoded
    • parsed
    • raw
    • standard
  • params
    • mimic3
    • mimic4
  • icd9.txt
• raw
  • mimic3
    • ADMISSIONS.csv
    • DIAGNOSES_ICD.csv
    • PRESCRIPTIONS.csv
  • mimic4
    • admissions.csv
    • diagnoses_ics.csv
    • prescriptions.csv

(2) Preprocess datasets

For the MIMIC-III dataset: 

```bash
python run_preprocess_mimic-iii.py
```

For the MIMIC-IV dataset: 

```
python run_preprocess_mimic-iv.py
```

2. Configuration

Please see train_hyper.py for detailed configurations.

3. Train model

You can train the model with the following command:

```bash
python train_hyper.py
```

Acknowledgement

If this work is useful in your research, please cite our paper.

@inproceedings{zhang2024multi, 
  title={Multi-Channel Hypergraph Network for Sequential
Diagnosis Prediction in Healthcare},
  author={Zhang Xin, Peng Xueping, Guan Hongjiao, Zhao Long, Qiao Xinxiao, Lu Wenpeng},
  booktitle={Proceedings of the 27th International Conference on Computer Supported Cooperative Work in Design (CSCWD 2024},
  year={2024}
 organization={IEEE}
}