Advancing COVID-19 data classification and prediction: A fresh perspective from an ontological machine-learning algorithm

This research introduces a unified framework for preparing, analyzing, and predicting COVID-19 data patterns using an ontological approach. Leveraging ontology models as a knowledge base, our framework enables more intelligent data analysis than existing state-of-the-art approaches. We also integrate two new concepts into the framework: a semantic decision tree that computes the involved information gain in decision tree construction, thereby improving the classification performance, and a method of autoregressive integrated moving average with exogenous semantic variables that forecasts the number of COVID-19 cases. This method is seamlessly integrated into the knowledge base to enhance the predictive power of the traditional approach such as autoregressive integrated moving average with explanatory variable. The core of our system is the COVID-19 knowledge base, which extracts the relevant data and leverages metadata for effective analysis and pattern learning. Experimental results demonstrate the superiority of our contributions over those corresponding to baseline methods. The higher accuracy and lower error rates of our approach than those of comparable methods are demonstrated using various criteria: mean square error, mean absolute error, root mean square error, area under the receiver operating characteristic curve, and classification accuracy.