A data-driven epidemiological model to explain the Covid-19 pandemic in multiple countries and help in choosing mitigation strategies

Abstract

Accurate models are fundamental to understand the dynamics of the COVID-19 pandemic and to evaluate different mitigation strategies. Here, we present a multi-compartmental model that fits the epidemiological data for eleven countries, despite the reduced number of fitting parameters. This model consistently explains the data for the daily infected, recovered, and dead over the first six months of the pandemic. The good quality of the fits makes it possible to explore different scenarios and evaluate the impact of both individual and collective behaviors and government- level decisions to mitigate the epidemic. We identify robust alternatives to lockdown, such as self- protection measures, and massive testing. Furthermore, communication and risk perception are fundamental to modulate the success of different strategies. The fitting/simulation tool is publicly available for use and test of other models, allowing for comparisons between different underlying assumptions, mitigation measures, and policy recommendations.

Modeling the COVID-19 Dynamics and Explaining the Portuguese CaseA Contribution for Fiting Differential Equation Solutions to Epidemiological Data

“…The main difference of this modified PSEIRD(S) model with respect to the original PSEIRD(S) [11] is the elimination of the infection pathway by which a fraction of the infectious individuals would not infect anyone else. Additionally, it was further assumed that the deceased came exclusively from the detected infected compartment, I d , as opposed to what was considered for the original PSEIRD(S) model.…”

“…This compartment is the one that is most subject to reporting delays and, if the reporting delays are not systematic, the data present several discontinuities that are difficult to incorporate in the model. We chose to attribute 14 days to this characteristic recovery time, based on the what was observed for other countries [11].…”

“…The number of daily new cases, deceased and recovered related to COVID-19 is made available to the general public in close to real time [15,16], which makes it possible to test models by actually fitting the existing the existing data. Here we propose a model based on the comprehensive PSEIRD(S) model by Beira et al [11] to fit the data sets related to the infected, deceased and hospitalized cases reported for Portugal, a country that had a severe post-Christmas outbreak. As fitting the solutions of differential equations is a such a non-trivial process and requires an extensive overhead work to implement dedicated software, an online open-access platform was developed.…”

“…The well-known works of W. O. Kermack and A. G. McKendrick [3], published in 1927, on the compartmental SIR model, have paved the way for epidemiologists to mathematically describe the dynamics of infectious diseases and became increasingly popular among the scientific community, particularly for the analysis of the current pandemic [4][5][6][7][8][9][10][11][12]. These compartmental SIRmodels are composed of systems of differential equations having no analytical solutions.…”

“…These compartmental SIRmodels are composed of systems of differential equations having no analytical solutions. Probably, this is the main reason why only a very limited number of scientific works actually provide the fit of the SIR model to the COVID-19 related data [6,7,11] and why others prefer to fit the data using approximated analytical expressions, such as the logistic function [13,14].…”