Skip to content
master
Go to file
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
 
 
 
 
 
 
 
 
 
 
 
 

README.md

VetBERT


Brian Hur, b.hur@unimelb.edu.au

VetBERT is a contextualized language model pretrained on over 15 million veterinary clinical notes and can be trained to perform a variety of tasks such as the disease indicated in a veterinary clinical record.

The classifier model implements VetBERT as described in the paper from BioNLP workshop @ ACL 2020 which can be used to classify the disease syndrome in a veterinary clinical note.

Instructions

To run, install the requirements

Download the zipped VETBERT model here

Download the zipped trained classifier here

unzip the folders contained in the files in the same file that the scripts are being ran.

ensure you have python 3.6 or higher running.


pip install requirements.txt

to perform test classification run:


python vetbert_classify_demo.py ./input/clinical_notes.xls

If test successful, you should see the output results and there should be a file in the folder:

./output/predicted_outputs.xls

To classify your own notes, follow the format in ./input/clinical_notes.xls and save using Excel 97-2003 format. You need to supply a dummy label if you do not have the labels and are note testing the model. The labels that can be used are listed in labels.txt.

The following paper should be cited if you use any of these resources:


@inproceedings{hur2020domain,
  title={Domain Adaptation and Instance Selection for Disease Syndrome Classification over Veterinary Clinical Notes},
  author={Hur, Brian and Baldwin, Timothy and Verspoor, Karin and Hardefeldt, Laura and Gilkerson, James},
  booktitle={Proceedings of the 19th SIGBioMed Workshop on Biomedical Language Processing},
  pages={156--166},
  year={2020}
}

Please comment or message me if you have any questions or run into any issues.

About

No description, website, or topics provided.

Resources

Releases

No releases published

Packages

No packages published
You can’t perform that action at this time.