Automating Biomedical Literature Review for Rapid Drug Discovery: Leveraging GPT-4 to Expedite Pandemic Response.
This repository contains the code and resources for our research project aimed at curtailing the time and resources traditionally associated with manual reviews in drug discovery and repurposing. Our automated framework identifies papers with potential drug targets, aiding researchers in maintaining an up-to-date map of current readiness for priority pathogens.
To set up the virtual environment and install the required libraries, execute the following commands:
conda env create -f environment.yml
conda activate drug
Before using the scripts in this repository, you need to obtain an API key from OpenAI and replace your API key in the scripts.
Note: Keep your API key secure and do not share it publicly.
This section provides detailed instructions on how to run the scripts in this repository.
python data_preparation/PubMed_abstract_extraction.py
python data_preparation/concatenate_sections.py
python data_preparation/generate_explanation.py
python data_preparation/generate_embedding.py
python data_preparation/run_cross_validation.py
python data_preparation/run_cross_validation_embeddings.py
python model/run_zero.py --cot 0
python model/run_zero.py --cot 1
python model/run_few_random.py --cot 0
python model/run_few_random.py --cot 1
python model/run_few_similar.py --cot 0
python model/run_few_similar.py --cot 1
Title: Automating Biomedical Literature Review for Rapid Drug Discovery: Leveraging GPT-4 to Expedite Pandemic Response
Authors:
- Jingmei Yang
- Kenji C. Walker
- Ayse A. Bekar-Cesaretli
- Boran Hao
- Nahid Bhadelia, M.D.
- Diane Joseph-McCarthy, Ph.D.
- Ioannis Ch. Paschalidis, Ph.D
title={Automating biomedical literature review for rapid drug discovery: Leveraging GPT-4 to expedite pandemic response},
author={Yang, Jingmei and Walker, Kenji C and Bekar-Cesaretli, Ayse A and Hao, Boran and Bhadelia, Nahid and Joseph-McCarthy, Diane and Paschalidis, Ioannis Ch},
journal={International Journal of Medical Informatics},
pages={105500},
year={2024},
publisher={Elsevier}
}
This project is open source and available under the MIT License.
- Email: [email protected].