COVID-19 Confirmed Infection Growth Prediction with Non-Pharmaceutical Interventions and Cultural Dimensions

The released publication of this work may be found here:

• Machine Learning–Based Prediction of Growth in Confirmed COVID-19 Infection Cases in 114 Countries Using Metrics of Nonpharmaceutical Interventions and Cultural Dimensions: Model Development and Validation (Journal of Medical Internet Research, 2021).

The main pipeline used in this study is in Experiment.ipynb. The library ml_pipeline.py contains functions used in the pipeline.

There are two directories:

• ./data/ contains the 3 data sets used in this study
• ./figures/ contains the figures shown in the publication. Running Experiment.ipynb will generate figures in this directory

Run

Running Experiment.ipynb will generate the tables and figures shown in the publication. Settings of the study may be modified in cell 4 of this notebook. These settings include:

• time_series_split_method: True runs the out-of-distribution validation method, False runs the in-distribution validation method. The country-based cross-validation method is ran additionally, regardless of the value of this parameter.
• run_models: Boolean dictionary indicating which models to run. True includes the model in the experiment. False excludes the model from the experiment.
• times_new_roman: True generates figures with the Times New Roman font. False generates figures with the default font.

Acknowledgements

This work is co-authored by Arnold YS Yeung, Francois Roewer-Despres, Laura Rosella, and Frank Rudzicz.