Identifying Country-Level Risk Factors for the Spread of COVID-19 in Europe Using Machine Learning.

Area of Science:

• Epidemiology
• Public Health
• Data Science

Background:

• Coronavirus disease 2019 (COVID-19) caused millions of deaths globally, impacting daily life and economies.
• Limited understanding exists regarding country-level factors influencing COVID-19 transmission and fatality within the European Union (EU).

Purpose of the Study:

• To identify country-level risk factors affecting COVID-19 transmission and fatality in the EU.
• To apply a machine learning (ML) predictive pipeline and explainability analysis for risk factor identification.

Main Methods:

• A hybrid dataset was created using public sources, including mobility, policy responses, vaccinations, and demographics for EU countries.
• Data pre-processing and feature dimensionality reduction were performed.
• A linear ε-Support Vector Machine (ε-SVM) model predicted COVID-19 deaths across three pandemic waves.
• Post hoc explainability analysis was conducted to understand ML decision-making.

Main Results:

• The ε-SVM model achieved a mean square error of 0.027 for wave 1 and less than 0.02 for waves 2 and 3 in predicting COVID-19 deaths.
• Explainability analysis revealed the contribution of selected country-level parameters to the prediction of COVID-19 deaths.

Conclusions:

• Machine learning effectively identified key country-level factors influencing COVID-19 deaths in the EU.
• The findings enhance the understanding of drivers behind COVID-19 transmission and fatality at a national level within the EU.