Prepare

• run prepare.py in real folder
• run gen_sim_data.py in simulation folder

Run Experiments

• run expert model (please run expert model first)
  • python main.py --exp expert
• run neural model
  • python main.py --exp neural
• neural model with augmented data
  • python main.py --exp data_aug
• tune expert model on the real-world data
  • python main.py --exp union

Train with partial data

You may use --frac augmentation to train the model with partial data: use --frac 0.1 to random sample 10% data; use --frac 540 to filter out the patient 540.

BG-ODE