This project mainly focus on the validation the impact of distribution shift on the risk prediction of major CVD event, here are the brief instructions below
- requirement.txt includes Python packages to run this project
- The folder "Jupyer" includes the training script for CPH models
- the yaml file in the example_config folder provides necessary parameters to train the BEHRT model
- when the yaml file is ready, run the main.py to train the BEHRT model, save_path is the directory to save log and model checkpoint
- python /main.py --params exmaple_config/bert.yaml --save_path ./