Quadrilogistic model applied to covid-19 deaths in Brazil, Minas Gerais and two of its cities

Abstract

The behavior of death cases accumulated in the COVID-19 pandemic suggests several phases of growth, which begin more slowly, go through a high growth rate and end with a tendency towards stability. Therefore, the main objective of this work is to model and analyze the quality of the fit of these data using the quadrilogistic growth model, considering four waves of growth in the number of deaths. The analysis was carried out for Brazil, the state of Minas Gerais, and the cities of Divinópolis and Formiga, in the Center-West of Minas Gerais. Initial estimates for the parameters were made based on official data and optimized using the Generalized Reduced Gradient method. The results showed that Minas Gerais had the lowest mean absolute percentage error and Brazil had the highest. In addition, all the regions analyzed had a high coefficient of determination, showing that all the model projections and official data have a robust linear correlation. It can be concluded that the model used significantly captured the growth dynamics of the number of deaths accumulated in the pandemic, including equilibrium and inflection points.