Predicting Local Surges in COVID-19 Hospitalizations

Dependencies

Packages: deampy (1.2.0), apacepy (1.0.15), pydotplus (2.0.2), and imblearn (0.0).

To create decision rules:

1. Calibrate the simulation model by running the script calibrate.py. This script will identify a set of simulated trajectories that can be used to develop and validate the decision rules.
2. Build all datasets for developing and validating the decision rules by running build_all_datasets.py. This scrip uses the simulated trajectories identified by the calibration procedure to create the dataset needed to develop and validate the decision rules.
3. Build and validate decision trees by running build_and_validate_decision_trees.py.
   1. The decision rules will be stored under outputs/figures/trees_4_weeks and outputs/figures/trees_8_weeks.
   2. The performance of decision rules under different scenarios will be stored under outputs/prediction_summary_4_weeks/dec_tree/summary.csv and outputs/prediction_summary_8_weeks/dec_tree/summary.csv.
4. If you would like to create a pruner decision trees, use build_a_decision_tree.py with a higher value of CCP_ALPHA.

Other functions:

• Run simulate_many.py to generate simulated trajectories, which will be stored in outputs/trajectories. Figure visualizing these trajectories will be stored under outputs/figures.