Abstract

We recently published a macroscopic virus-centric model to simulate the evolution of the SARS CoV 2 epidemic (COVID 19) at the level of a country or a geographical entity using a new decomposition modeling and machine learning optimization. The approach assumes that an epidemic can be analyzed as the combination of several elementary epidemics representing each different parts of the population with different behaviors through time, different locations, or different phases of the virus propagation like emergence of new variants. In part 1 of the paper, published in 2020, we presented the details of the model and its application for new cases through different countries. In this second part, we develop and analyze an application of this modeling to the number of deceased cases among 22 different countries in Europe. The proposed modeling is still based on the finite superposition of Verhulst equations commonly known as logistic functions and used in population dynamics. The novelty comes from the new decomposition of a complex event and the use of machine learning algorithm. The results show that this approach enables to well simulate the evolution of the number of deaths for the different analyzed countries, population, or age. It also shows that the epidemic kinetic can be well simulated whether you consider the overall epidemic kinetic as one epidemic or as the sum of independent epidemics, as is presented here regarding the age of the population. The modeling prediction quality was also studied as a function of the amount of available data.

Keywords

Mathematical Subject Classification

Milestones

Authors