Neural Network powered COVID-19 spread forecasting model

Wieczorek, Michał; Siłka, Jakub; Woźniak, Marcin

Virus spread prediction is very important to actively plan actions. Viruses are unfortunately not easy to control, since speed and reach of spread depends on many factors from environmental to social ones. In this article we present research results on developing Neural Network model for COVID-19 spread prediction. Our predictor is based on classic approach with deep architecture which learns by using NAdam training model. For the training we have used official data from governmental and open repositories. Results of prediction are done for countries but also regions to provide possibly wide spectrum of values about predicted COVID-19 spread. Results of the proposed model show high accuracy, which in some cases reaches above 99%.