Code for the paper Dynamically Extracting Outcome-Specific Problem Lists from Clinical Notes with Guided Multi-Headed Attention. This repo is currently a work in progress as I continue to clean up the code. Feel free to create an issue if there are any problems.
Set the appropriate data paths in resources/config.yml.
python extract_labels.py
python cross_val_splits.py
python word_embeddings.py
Set the training hyperparameters and define the task in resources/params.json.
python train.py
We adapted publicly available code from the following sources:
- Preprocessing and modeling code from the DeepEHR repo
- Modeling code from the CAML repo
- Boilerplate code (e.g. loading data, training) from Stanford's CS230 Code Examples repo