BJH-perioperative-notes-bioGPT

This clinical foundational model is intended to predict post-operative surgical outcomes from clinical notes taken during perioperative care. It was finetuned from the microsoft/biogpt model through a multi-task learning approach, spanning the following 6 outcomes:

• Death in 30 days
• Deep vein thrombosis (DVT)
• pulmonary embolism (PE)
• Pneumonia
• Acute Knee Injury
• delirium

Also check out cja5553/BJH-perioperative-notes-bioClinicalBERT, which is the bioGPT variant of our model!

Dataset

We used 84,875 perioperative clinical notes from patients spanning the Barnes Jewish Healthcare (BJH) system in St Louis, MO. The following are the characteristics for the data:

• vocabulary size: 3203
• averaging words per clinical note: 8.9 words
• all single sentenced clinical notes

How to use model

from transformers import BioGptTokenizer, AutoModelForCausalLM
model=AutoModelForCausalLM.from_pretrained("cja5553/BJH-perioperative-notes-bioGPT")
tokenizer = BioGptTokenizer.from_pretrained("microsoft/biogpt")

Codes

Codes used to train the model are publicly available at: https://github.com/cja5553/LLMs_in_perioperative_care

Citation

If you find this model useful, please cite the following paper:

@article{
author={Bing Xue, Charles Alba, Joanna Abraham, Thomas Kannampallil, Christopher King, Michael Avidan, Chenyang Lu}
title={"Prescribing Large Language Models for Perioperative Care: What’s The Right Dose for Pretrained Models?"},
year={2024}
}

Questions?

contact me at alba@wustl.edu