ADRNet: Multi-label Adverse Drug Reaction Prediction via Drug Descriptor-aware Collaborative Filtering.

Usage

• Create and activate a python environment using anaconda:

  conda env create -f py37env1.yml

  conda activate py37env1

  pip install torch==1.9.1+cu111 torchvision==0.10.1+cu111 torchaudio==0.9.1 -f https://download.pytorch.org/whl/torch_stable.html

• To generate K-Fold data:

  python main.py -i

• To run and evaluate a model:

  python main.py -d DATA_NAME -m MODEL_NAME -f FEATURE_TYPE

  For example: python main.py -d AEOLUS -m DrugNCF -f 2

  Evaluation results containing AUC, AUPR and STDERR are stored in "./final_results" folder.

• To obtain options for DATA_NAME and MODEL_NAME and FEATURE_TYPE:

  python main.py -h

Data

All input data is available in the "./data" folder:

Citation

If you find our codes are helpful, please kindly cite our paper below:

@inproceedings{li2023adr,
  title={A{D}{R}{N}et: A Generalized Collaborative Filtering Framework Combining Clinical and Non-Clinical Data for Adverse Drug Reaction Prediction},
  author={Li, Haoxuan and Hu, Taojun and Xiong, Zetong and Zheng, Chunyuan and Feng, Fuli and He, Xiangnan and Zhou, Xiao-Hua},
  booktitle={ACM Conference on Recommender Systems},
  year={2023}
}

Reference

@article{nguyen2021survey,
  title={A survey on adverse drug reaction studies: data, tasks and machine learning methods},
  author={Nguyen, Duc Anh and Nguyen, Canh Hao and Mamitsuka, Hiroshi},
  journal={Briefings in Bioinformatics},
  volume={22},
  number={1},
  pages={164--177},
  year={2021},
  publisher={Oxford University Press}
}