COVID-19 Severity Analysis Using Improved Machine Learning Algorithm

The new pandemic produced by the COVID-19 virus has resulted in an overflow of medical treatment in clinical centers all over the world. The fast and exponential growth in the number of COVID-19-infected individuals has necessitated an effective and timely prediction of probable infections and their effects in order to reduce health-care quality overload. As a result, intelligent models are being developed and used to assist medical workers in making more accurate diagnoses concerning the health condition of COVID-19-infected individuals. The purpose of this research is to present an alternative algorithmic approach for predicting the health status of COVID-19 patients in Mexico. Different prediction models were assessed and compared, including Adaboost, gradient boosting machine, random forests, and light gradient boosting machine. Additionally, Grid search hyperparameter optimization is used to improve the algorithm's success rate. The optimal model feature analysis procedure is being carried out. The purpose of this study is to analyses features in terms of feature importance as indicated by SHapely adaptive exPlanations (SHAP) values in order to identify relevant predictive factors that can identify patients at high risk of mortality.