This repository provides the code to run a disparity analysis using process mining techniques. It is applied to patients within MIMIC-IV that have sepsis but can be extended to any other event log. To extract the event log of sepsis patients from MIMIC please
- Apply the sql statements in mimic_extraction/ to MIMIC-IV and store the resulting tables.
- Apply the Full_preprocessing_notebook.ipynb to retrieve the event log.
After that, you can run the analysis on the event log. You can do so by:
- Running the main.py script to perform the hypothesis test on the distribution of variants across patient attributes and to generate the aggregated variants with timing and SOFA information.
- Applying the statistics.py script to generate descriptive statistics and testing for their distribution across patient groups.
- Applying the visualization.py script to generate the visualizations of variants, along with the aggregated timing and SOFA information for all patient groups.