Predicting COVID-19 Mortality Rates and Identifying High-Risk Areas Using Machine Learning Models

Abstract

The COVID-19 pandemic has wreaked havoc on global society, exerting unprecedented strain on national healthcare systems. This public health catastrophe has compelled governments and communities to reassess their preparedness for health crises. As of April 2024, the virus has reached 181 countries, with more than 120,000 confirmed cases reported. Research employing various machine learning models—including random forest, LASSO, and multiple regression—has been conducted to predict the mortality rate of the COVID-19 pandemic. These models consider multiple factors, such as the rate of confirmed cases, the number of hospital beds per thousand people, the duration of infection, the total vaccinated population, and the median age in each country. Initial findings indicate that the rate of confirmed cases, hospital bed availability, and the duration of infection are the most significant predictors of mortality rates. Based on these models, countries have been categorized into low- and high-risk groups. Consequently, the capacity of clinical facilities in terms of beds per thousand people can also be forecasted, enhancing readiness for potential future outbreaks. This proactive approach could offer critical insights for governmental and health organizations, potentially mitigating the impact of subsequent COVID-19 waves.