"RASNet: Recurrent Aggregation Neural Network for Safe and Efficient Drug Recommendation"

Results

Results of the paper:

DDI: 0.0599(0.0009) Ja: 0.5401 (0.0021) PRAUC: 0.7882 (0.0025) F1: 0.6931 (0.0019)

Folder Specification

data/output/
- ddi_A_final.pkl: ddi adjacency matrix
- ehr_adj_final.pkl: used in GAMENet baseline (if two drugs appear in one set, then they are connected)
- records_final.pkl: The final diagnosis-procedure-medication EHR records of each patient
- voc_final.pkl: diag/prod/med index to code dictionary
src/
- model.py: Code for model definition.
- util.py: Code for metric calculations and some data preparation.
- test.py: Code for reproducing the paper results
- layer.py

We cannot distribute the whole MIMIC-III data https://physionet.org/content/mimiciii/1.4/, then please download the dataset by yourself.

Package Dependency

first, install the conda environment

conda create -n RASNet python=3.8
conda activate RASNet

then, in RASNet environment, install the following package
```
pip install scikit-learn, dill
```

Finally, install other packages if necessary

pip install [xxx]

Here is a list of reference versions for all package

pandas: 1.5.2
dill: 0.3.6
torch: 2.0.1
scikit-learn: 1.2.0
numpy: 1.23.4

Run

Run test.py to reproduce the results of the paper.

cd src
python test.py

Citation

If the code and the paper are useful for you, it is appreciable to cite our paper:

@article{zhu2024rasnet,
  title={RASNet: Recurrent aggregation neural network for safe and efficient drug recommendation},
  author={Zhu, Qiang and Han, Feng and Yang, Huali and Liu, Junping and Hu, Xinrong and Wang, Bangchao},
  journal={Knowledge-Based Systems},
  pages={112055},
  year={2024},
  publisher={Elsevier}
}

Partial credit to previous reprostories:

cnhfeng/RASNet

"RASNet: Recurrent Aggregation Neural Network for Safe and Efficient Drug Recommendation"

Results

Folder Specification

Package Dependency

Run

Citation