ABSTRACT
The early classification of COVID-19 patients severity can help save lives by giving to doctors valuable instructions and guidelines for the cases that may need more attention to survive. This paper aims to classify cases depending on their severity into three classes: "survivor”, "sudden death” and "death” using electronic health records (HER). The first class represents positive cases discharged from the hospital after being treated for COVID-19. While the second and the third classes are describing the level of cases severity based on the interval of death. We called the highest severity class "sudden death” to identify critical cases with a high risk of death in the first two days of admission, while the "death” class includes severe cases with an interval of death beyond two days. The sudden death class represents the biggest challenge for this classification as the number of samples representing this case is very small. This paper presents a triage system for COVID-19 cases using four machine learning algorithms (KNN, Logistic Regression, SVM, and Decision tree). The best classification results were obtained using Logistic Regression and SVM models. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.