Supervised Machine Learning Model For Effective Classification Of Patients With Covid-19 Symptoms Based On Bayesian Belief Network

Covid-19 is a contagious infection and should be managed properly. If it is identified early enough, the chances of survival are high. Symptoms presentations may be confusing to health practitioners since they are similar to other diseases and the number of health professionals are mostly insufficient, especially in developing countries, hence a correct and timely diagnoses based on symptoms’ presentations can prove difficult. So far, efforts to take address these needs are still not adequate. In this paper, we leverage on the efficacy of machine learning algorithms to present a machine learning approach, Bayesian Belief Network, that aims at ensuring that the symptoms’ set are correctly and timeously classified as either Covid-19+ or Covid-19- . A dataset is gathered from observed symptoms of confirmed Covid-19 patients by healthcare practitioners. The dataset is trained so as to be able to classify, based on this prior knowledge, any supplied symptom set. This has been able to effectively handle the problem of wrong or delayed diagnoses with its attendant negative consequences. Our model yields 98% accuracy, showing a significant classification accuracy of patients with Covid-19 symptoms.