ABSTRACT
The world faces a rapidly spreading of COVID19 globally, for several countries around the world mitigating the consequences and spread of the pandemic remains a top priority. Researchers work to find a smart and rational solution to limit the spread of this epidemic and its repercussions. The goal of this research is to produce an early and accurate COVID-19 prediction, as well as a comparative analysis of the performance of several machine learning (ML) models based on patient vital signs, dataset balancing, and feature selection. The cases dataset was provided by King Fahad Hospital University in Al-Khobar, Saudi Arabia. The current study used the WEKA 3.8.5 and Python programming language (SKLEARN) to decide which method generated the highest level of accuracy while using fewer features. Random forest with grid search (RF with grid search), Artificial Neural Networks (ANN), Support Vector Machine (SVM), Random Forest (RF), J48, XGB Classifier, and XGB Classifier with grid search were the techniques that were compared. The highest level of accuracy obtained with seven features was 84% achieved with the RF using grid search technique, while ANN, SVM, RF, J48, XGB Classifier, and XGB Classifier with grid search obtained 82.85%, 79%, 82.93%, 82.5%,82.21%, and 83.4% accuracy, respectively. © 2022 IEEE.