ABSTRACT
The deadly disease SARS-CoV-2 caused an outbreak of the emerging infectious ailment of COVID-19. The different pathogenies have been categorized as variants of concern. The Omicron variant had a higher affinity and caused more asymptomatic infections than the other. Identifying and preventing COVID-19 in addition to combating the fast spreading of the disease at an early stage is imperative. This paper proposes a CoV-3D model for deadly diseases (COVID-19). It uses the ensemble machine learning (SMOTE) technique with laboratory blood test parameters. The model has been developed to predict the deadly disease from standard blood parameters: neutrophil count, complete lymphocyte count, and hematocrit. The proposed classical model has been created, tested, and evaluated using an open-source blood test data set. The first stage classifier used a state-of-the-art supervised learning algorithm. The subsequent phase used K-Nearest Neighbors, Support Vector Machine, Naive Bayes, Decision Tree, Random Forest, and Logistic Regression. The proposed model employs a synthetic minority oversampling technique to balance a data distribution. The proposed method attains an excellent prediction accuracy of 0.80%, the AUC (0.84%), and F1 Score (0.80%).
