Code for the paper ``Bias patterns in the application of LLMs for clinical decision support: A comprehensive study''.

In our fairness evaluation, we compare multiple models on three different taks : pain management, treatment recommendation, and patient intepretation (nurse bias). Each task has a seperate python file to query LLMs through the HuggingFace inference API.

If you find our work useful, please consider the following citation:

@misc{poulain2024bias,
      title={Bias patterns in the application of LLMs for clinical decision support: A comprehensive study}, 
      author={Raphael Poulain and Hamed Fayyaz and Rahmatollah Beheshti},
      year={2024},
      eprint={2404.15149},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}