Classification of patiens whit COVID-19 predisposed to intensive care using SVM and RANDOM FOREST techniques

Abstract

COVID-19 is a highly contagious respiratory disease caused by SARS-CoV-2. Predictive algorithms would allow the identification of those who could be admitted to intensive care. In this work, a methodology was followed that consists of selecting a data set, which will then be processed through techniques such as One Hot Encoding, MICE, and LASSO. Then the proposed Random Forest and Support Vector Machine models will be developed and evaluated using sensitivity, specificity, and AUC metrics. The results indicate that the Random
Forest model obtains a better performance for the classification of patients who are admitted to an intensive care ward.