Mathematical Modeling of Covid-19 Spread Using Genetic Programming Algorithm

Abstract

This paper analyses the possibilities of using Machine learning to develop a forecasting model for COVID-19 with a publicly available dataset from the Johns Hopkins University COVID-19 Data Repository with the addition of the percentage of each variant from the GISAID Variant database. The Genetic programming (GP) symbolic regressor algorithm is used for the estimation of new confirmed cases,hospitalized cases,cases in intensive care units (ICUs),and the number of deaths. This metaheuristics method algorithm is made from a dataset for Austria and its neighboring countries the Czech Republic,Slovenia,and Slovakia. Machine learning was performed twice to create individual models for each country,but the second time the process covered all countries at once as a multi-country model. Variance-based sensitivity analysis was initiated using the obtained mathematical models. This analysis showed us on which input variables the output of the obtained models is sensitive,like in case of how much each covid variant affects the spreading of the virus or the number of deaths. Individual short- term models show very high R2 scores,while long-term predictions have lower R2 scores. The multi-country model achieved inferior results as additional valuables needed to be added in order to obtain better results.