A Bayesian Network Model for the Prognosis of the Novel Coronavirus (COVID-19)

Aliyu, S.; Zakari, A. S.; Adeyanju, I.; Ajoge, N. S.

The World Health Organization (WHO) classified the new coronavirus disease 2019 (COVID-19) as a Public Health Emergency of International Concern on January 31, 2020. It is now clear that the dreadful virus has put a significant burden on the world's healthcare systems. Currently, the main techniques for diagnosing COVID-19 are viral nucleic acid testing and chest computed tomography (CT). Though proven to be effective, these methods are time consuming hence, the need for the use of non-clinical approaches for early detection, diagnosis and prognosis of the coronavirus. Using epidemiological dataset of COVID-19 patients, this study presents the use of Bayesian network model to predict contraction of the coronavirus disease. Following the application of several structural learning techniques, a causal Bayesian network based on the nodes: breathing problem, fever, dry cough, sore throat, running nose, asthma, chronic lung disease, headache, heart disease, diabetes, hypertension, fatigue, gastrointestinal, abroad travel, contact with COVID Patient, attended Large Gathering, visited public exposed places, family working in public exposed places and COVID-19 status, was estimated. The developed BN model correctly predicted the probability of contracting COVID-19 with an accuracy of 98%. The model also affirms that, individuals with contact history with a COVID-19 patient are the most susceptible to contracting the coronavirus disease. © 2021, Springer Nature Switzerland AG.