Search

In December 2019, China reported a series of atypical pneumonia cases caused by a new Coronavirus, called COVID-19. In response to the rapid global dissemination of the virus, on the 11th of Mars, the World Health Organization (WHO) has declared the outbreak a pandemic. In light of this situation, this paper intends to analyze and improve the current SEIR models to better represent the behavior of the COVID-19 and accurately predict the outcome of the pandemic in a given social, economic and political scenario. We present a novel generalized Susceptible-Exposed-Infected-Recovered (SEIR) compartmental model and test it using a global optimization algorithm with data collected from the WHO. Our main results were: (a) our model was able to accurately fit the data of all countries tested (b) it is possible to predict values for one week ahead with errors in the order of 15% for the number of cases and 30% in the number of deaths for all countries; (c) predictions are better for countries where the active cases curve already reached the maximum; the error being in the order of 10% in the number of cases and 20% in the number of deaths; (d) for countries where the active curve is still growing, different optimization solutions can be found that fit the data; so, to predict future behavior in this scenarios some of the model coefficients should be estimated from outside sources or based on generalized results from other countries according to their health policies of social distance, quarantining and case test and tracing.

The objective of the current investigation was to produce a generalized computational model to predict consequences of various reopening scenarios on COVID-19 infections rates and available hospital resources in Sao Paulo - Brazil. We were able to use the Susceptible-Exposed-Infected-Recovered (SEIR) model to fit both accumulated death data and corrected accumulated cases data associated with COVID-19 for both Brazil and the state of Sao Paulo. In addition, we were able to simulate the consequences of reopening under different possible scenarios in Brazil, in special for the state of Sao Paulo. The model was able to provide a predicted scenario in which reopening could occur with minimal impact on human life considering people careful behavior in combination with continued social distancing measures.