Before extracting data, one need to request permission and download MIMIC III dataset from https://mimic.physionet.org/gettingstarted/access/ as csv.gz files
Before loading data, “LABEVENTS.csv.gz”, “CHARTEVENTS.csv.gz”, “OUTPUTEVENTS.csv.gz” should be saved in a folder named raw_data, and all loaded items would be saved in the raw_data folder.
python3 load_data.py
Before cleaning data, a folder named “remove_outlier” need to be created.
python3 clean_data.py
Please refer to the following jupyter notebook files
- los.ipynb
- mortality.ipynb
Please follow and run the folloing jupyter notebook files in order
- process_spas.ipynb
- process_data.ipynb
- feature_selection.ipynb
Please follow the steps in the following jupyter notebook file
- NN.ipynb
The GRU model needs to take a parameter --target to specify predicting length of stay or mortality rate. It could be run like:
python3 interpolation_GRU.py --target 3
python3 interpolation_GRU.py --target 7
python3 interpolation_GRU.py --target m