data

Raw Datasets

P12

• download the dataset from PhysioNet/Computing in Cardiology Challenge 2012. The dataset contains three folder for three hospitals. You can also download the csv file containing all 12k patients from here(p12.csv).

P19

• download the dataset from PhysioNet/Computing in Cardiology Challenge 2019. The dataset contains two folders for two hospitals. You can also download the csv file containing all ~43k patients from here(df_A.csv and df_B.csv).

MIMIC-III

• You need to request access to the dataset from PhysioNet. Once you downloaded the csv files, you can use this repo to extract ~50k patients into a .csv fuke.

Put the csv files in the data/raw folder or another folder. Then, add the path_raw to the configs/data/{DATASET_NAME}.yaml file.

Simulated data

• Please run:

  python gen_sim.py --n-vars 16 32 64 128 --lambdas 0.2 0.5 1 2
  python gen_sim.py --n-vars 16 --lambdas 0.5

  This will create 16 simulated datasets with different number of features and lambdas. The raw and processed datasets will be saved in the data/raw and data/processed folders, respectively. Create yaml file for the simulated data in the configs/data/{DATASET_NAME}.yaml.

Preparing Datasets

• Now, check Prepare_Datasets.ipynb to see how to prepare the raw datasets. The prepared datasets for P12 and P19 can be downloaded from here.