UML-Med-Framework-Reproducible-Package
Description
The package provides necessary Python notebooks, data files, intermediate and final output files, and instructions to train and test a CVD prediction model and use Uncertainty modeling for model selection and SHAP for model-agnostic feature importance and impact analysis. The README file provides detailed installation and run instructions. These are computational components of UML-Med Framework developed by Brahma et. al.
Files
Steps to reproduce
1. README.md is a mark-up file that contains descriptions of all files in the .zip file 2. .zip file: Contains subfolders - Data, Output, Program, Excel. After unzipping the programs can run based on this folder structure without modification of path in the code. When programs run successfully it overwrites the existing output files in the current folder with new results. 3. Refer to the Methodology or Data and methods sections in Brahma et. al articles related to CVD prediction 4. The input data file "cvd_ml_model_ready_dataset.csv" is a preprocessed data file that has been prepared from the original data files at source at https://sleepdata.org/datasets/shhs. To reproduce this research you need to use just this input data file. To access the original unprocessed dataset you might have to get necessary permissions from sleepdata.org