Publication

Abstract : Coronavirus Disease 2019 (COVID-19) has impacted our world. Risk classification during hospitalization is essential for medical planning and allocation of resources, and hence for lowering death rates, especially in underdeveloped countries. Many characteristics related to patients that influence illness severity, such as pre-existing comorbidities, can be employed to improve this prediction. Finding a biomarker which helps to identify individuals who need prompt treatment and determine their mortality risk has thus become a pressing yet difficult task. There is no advanced tool available for this. As a result, our research aims to develop a ML-based predictive model as well as a decision support system that can predict mortality based on clinical and health characteristics. We used different ML algorithms such as Random Forest, Support Vector Machines, XGBoost Classifier, and Logistic Regression for survival rate prediction in covid patients. Two feature selection methods, Information Gain and SVM-RFE were also incorporated. Accuracy comparison on 3 different datasets using the ML algorithms was done and the important biomarkers that helps in mortality prediction were identified.