clinical-assertion-data

This repository contains the assertion data introduced in our paper: Assertion Detection in Clinical Notes: Medical Language Models to the Rescue?, NLPMC @ NAACL 2021.

You can also try out how our best model performs in our demo application.

The model is also available on the Huggingface model hub: https://huggingface.co/bvanaken/clinical-assertion-negation-bert

Assertion Labels

We release 5,000 assertion labels manually annotated based on clinical notes from the MIMIC-III database. The table below shows the distribution of labels for the different note types.

	PRESENT	POSSIBLE	ABSENT
Discharge Summaries	2,610	250	980
Physician Letters	204	34	66
Nurse Letters	293	14	59
Radiology Reports	285	67	138

Creating sample files

In order to use our data, you need to have access to the MIMIC-III database, which is freely available after completing the steps described here: https://mimic.mit.edu/iv/access/.

To create training samples from our label files, simply follow these steps:

1. Install Requirements: pip install -r requirements.txt

2. Run conversion script:

python3 convert_to_samples.py \
 --label_path {...} \   # required, path to label file, e.g. labels/nursing_labels.csv
 --mimic_notes_path {...} \   # required, path to MIMIC-III NOTEEVENTS.csv
 --save_path {...}   # path to save created samples

3. Don't forget to add the special token [entity] to your model vocab for training.

Cite

@inproceedings{vanAken21,
    title = "Assertion Detection in Clinical Notes: Medical Language Models to the Rescue?",
    author = "van Aken, Betty  and
      Trajanovska, Ivana  and
      Siu, Amy  and
      Mayrdorfer, Manuel  and
      Budde, Klemens  and
      Loeser, Alexander",
    booktitle = "Proceedings of the Second Workshop on Natural Language Processing for Medical Conversations",
    year = "2021",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/2021.nlpmc-1.5"
}