Assessment of supervised machine learning methods for predicting COVID-19 likelihood

The rapid global spread of the SARS-CoV-2 virus, originating in China, created an unprecedented challenge for public health systems worldwide. Artificial intelligence (AI)-based predictive methods offer valuable tools for evaluating and shaping policy measures. These AI-driven techniques have garnered extensive interest from researchers employing machine learning and statistical models. This study explores a time-series-based prediction of COVID-19 cases using machine learning classifiers, including Naïve Bayes Classifier (NBC), Decision Tree Classifier (DTC), Linear Discriminant Analysis (LDA), Support Vector Machine (SVM), and Artificial Neural Network (ANN). Data were drawn from official health reports and academic publications. Notably, DTC and ANN achieved remarkable classification accuracies of 98.6% and 97.3%, respectively, while NBC also performed well with an accuracy of 94.5%. These findings indicate that DTC and ANN can better distinguish varying levels of SARS-CoV-2 infection risk compared to methods utilizing selected features as in NBC, SVM, and LDA.