LLM Benchmark for Clinical-Decision Support

This repository contains the code to create the curated dataset from the original MIMIC-IV-ED, MIMIC-IV and MIMIC-IV-Note datasets.

Generate the dataset from the MIMIC-IV datasets

Download the following datasets:
MIMIC-IV-ED: https://physionet.org/content/mimic-iv-ed/2.2/
MIMIC-IV: https://physionet.org/content/mimiciv/3.0/
MIMIC-IV-Note: https://physionet.org/content/mimic-iv-note/2.2/

Run the jupyter notebooks in the following order:
MIMIC-IV-Ext-Creation.ipynb
create_ground_truth_specialty.ipynb
Claude_triage_diagnosis_specialty.ipynb
postprocessing.ipynb

Run the .py files in the following order:
MIMIC-IV-Ext-Creation.py
create_ground_truth_specialty.py
CLaude_diag_spec_GeneralUser.py
CLaude_diag_spec_ClinicalUser.py
CLaude_triage_GeneralUser.py
CLaude_triage_ClinicalUser.py
postprocess_specialty_prediction.py
postprocess_diagnosis_prediction.py
postprocess_triage_prediction.py
additional_postprocess.py