A Deep Learning Approach for COVID-19 Infection Probability Prediction

Abstract

COVID-19 is one of the major health crises worldwide. Though number of vaccines is introduced by many countries but still it is challenging to detect the disease at early stage. Many advanced technologies are introduced for this purpose to stop the spread. Machine learning based COVID detection can be a supportive tool for both physicians as well as patients for early prediction of this illness. Different automated technologies are also found from the literature for this purpose. Deep learning approach is used for predicting the infection probability by analyzing the five types of COVID symptoms. The experiment is carried out with 2889 samples collected from a publicly available database. A deep neural network (DNN) classifier is designed for this purpose and the result is also compared with support vector machine (SVM). From the result it is observed that around 97% classification accuracy is observed with DNN classifier and it is better than SVM.