Proposing a Novel Artificial Neural Network Based Methodology for Forecasting Risk of Covid-19 Pandemic

Taha Osama A Binhuraib

Abstract: The corona virus formally known as Covid-19 has taken the world by storm. In this article we aim to analyze how harnessing the prowess of different computational methods –Machine Learning in particular; thus, helping policy makers take effective decisions by utilizing efficient approaches. Firstly, popularly known approaches (linear regression and logistic growth) are tried for existing data but it is figured out that accuracy rates are not satisfactory enough. Therefore, secondly, an artificial neural network (ANN) based methodology that consists of input data in different characters is proposed. It is figured out that the test results are more accurate with an R-squared score of (0.81) on an unseen data set, and handling the defined forecasting problem by using this multi-faceted data set is beneficial for policy makers, doctors and/or health managers that goal to have foresight on target groups at higher risk.