•  
  •  
 

Abstract

The coronavirus disease 2019 (COVID-19) has diverse impacts on human lives beyond being just a disease, causing mortality and morbidity. Existing studies have played a key role in curbing the disease’s progression. However, these studies are inconclusive because hospitals’ patients-to-bed ratio is way larger than that of beds-to-patients in some developing countries. In this study, I proposed and developed a novel decision support system that guides medical staff in identifying patients who need to be hospitalized as well as predicting the mortality caused by COVID-19. Five machine-learning models were tested: Decision Tree (DT), Random Forest(RF), Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and eXtreme Gradient Boosting (XGBoost). Two clinical datasets are processed using machine learning models with 5-fold cross-validation. The first dataset results showed the mean performance balanced accuracy of DT, KNN, SVM, RF, and XGBoost were 0.87, 0.87,0.59,0.87, and 0.87, respectively, and the mean area under the curve (AUC) was 0. 88, 0.87, 0.59, 0.89, and 0.89, respectively. The second dataset results showed the mean validation accuracy of DT was 0.926, and the mean AUC of DT was 0.947. The final decision was supported by the majority voting (Ensemble Algorithm-Hard Voting) of contributing models.

Keywords

Machine Learning Models, Decision Support System, COVID-19 Mortality, Majority Voting

Article Type

Article

First Page

74

Last Page

82

Publication Date

4-30-2025

Download

Share

COinS