Identifying novel factors associated with COVID-19 transmission and fatality using the machine learning approach

Abstract

The COVID-19 virus has infected more than 38 million people and resulted in more than one million deaths worldwide as of October 14, 2020. By using the logistic regression model, we identified novel critical factors associated with COVID19 cases, death, and case fatality rates in 154 countries and in the 50 U.S. states. Among numerous factors associated with COVID-19 risk, economic inequality enhanced the risk of COVID-19 transmission. The per capita hospital beds correlated negatively with COVID-19 deaths. Blood types B and AB were protective factors for COVID-19 risk, while blood type A was a risk factor. The prevalence of HIV and influenza and pneumonia was associated with reduced COVID-19 risk. Increased intake of vegetables, edible oil, protein, vitamin D, and vitamin K was associated with reduced COVID-19 risk, while increased intake of alcohol was associated with increased COVID-19 risk. Other factors included age, sex, temperature, humidity, social distancing, smoking, health investment, urbanization level, and race. High temperature is a more compelling factor mitigating COVID-19 transmission than low temperature. Our comprehensive identification of the factors affecting COVID-19 transmission and fatality may provide new insights into the COVID-19 pandemic and advise effective strategies for preventing and migrating COVID-19 spread.

Keywords: COVID-19 fatality; COVID-19 transmission; Machine learning; Protective factor; Risk factor.