A MULTINOMIAL NAÏVE BAYES DECISION SUPPORT SYSTEM FOR COVID-19 DETECTION

Abstract

Coronavirus disease 2019 termed COVID-19 is a highly infectious and pathogenic illness caused by severe acute respiratory syndrome. Symptoms of COVID-19 range from mild to severe, in some cases leading to death. Early detection could help to monitor progression of the disease, mitigate spread of the disease and possibly reduce mortality rate. Computer-aided diagnosis systems are designed to complement health care systems and assist in the early detection of diseases. Currently, as it is not possible to test all citizens especially in developing countries with very large populations due to financial constraints and the standard of their healthcare facilities, the problem of identifying suspected cases and deciding laboratory test priority among citizens is evident and more pressing. Therefore in this study, we introduce an interactive Artificial Intelligent web system using the Multinomial Naïve Bayes algorithm with the aim of detecting warning COVID-19 symptoms and to provide fitting suggestions.  Furthermore, the study also evaluates the performance of the Multinomial Naïve Bayes based on the different holdout approaches experimented. The experimental results are promising as the Multinomial Naïve Bayes is shown to achieve high accuracy detection thus providing a reliable method to identify warning symptoms of COVID-19.