A Time Series Analysis and Predictive Modeling of COVID-19 Impacts in the African American Community

Abstract

ABSTRACTBackgroundSometimes in 2019, there was an outbreak of coronavirus pandemic. Data shows that the virus has infected millions of people and claimed thousands of lives. Vaccination and other non-pharmacological interventions have brought a relief; however, COVID-19 left some indelible marks. This work focuses on a time series analysis and prediction of COVID-19 fatality rates in the Black community. Decision makers will find the work useful in building a robust architecture for a resilient pandemic preparedness and responsiveness against the next pandemic. Method: Our analysis of COVID-19 cases and deaths spans March 2020 to December 2020. Assuming there was no vaccine and other factors remained the same, we hypothesized that COVID-19 disproportionality would have continued. To test our hypothesis, COVID-19 forecasting cases and deaths models were built for the total population as well as the Black population. Holt and Holt-Winters exponential smoothing forecast methodologies were used for the forecast modeling. Forecasting accuracy was based on Mean Absolute Percentage Error (MAPE). Furthermore, we designed, developed, and evaluated a fatality rate predictive model for a Black county. Considering the number of ethnic groups in the USA, a Black county was defined as any county in the USA that at least 45% of its population are Blacks. Five learning algorithms were trained and evaluated. Dataset was a merger of datasets obtained from John Hopkins COVID-19 repository, US Census Bureau and US Center for Disease Control and Prevention.Results and ConclusionTime series analysis shows that there exists a strong evidence of COVID-19 disproportionate impacts in the states investigated. Using 9 different criteria for performance comparison, our predictive modeling showed that decision tree model has a slight edge over other models. Our experiment suggests that Blacks and senior citizens with pre-existing condition living in Georgia State are the most vulnerable to COVID-19.