Early predictive model for COVID-19 mortality and ventilator support: reducing risks for patients

Abstract

The COVID-19 virus, which is caused by the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), has the potential to generate a wide range of symptoms, including mild respiratory symptoms, severe respiratory distress, and even death. Prediction models cannot replace clinical judgment and patient assessment. Instead of depending solely on the estimates provided by models, healthcare providers should use them as decision-support tools to help them make the best possible choices. As part of this investigation, we construct a prediction model that we refer to as the Hybrid Spider Monkey Optimized eXtreme Gradient Boosting (HSMO-XGB). Patients who were part of the COVID-19 trial were hospitalized in King Abdul-Aziz Medical City in Riyadh for the purpose of this study. The dataset includes the patient's personal data, the outcomes of any tests performed in laboratories, and any chest X-ray (CXR) outcomes. The first step in preprocessing is called normalization, and the second step is called feature extraction using linear discriminate analysis (LDA). The reasonable expectation model for HSMO-XGB that is presently being shown will work on the administration of clinics by helping doctors in the early recognizable proof of Coronavirus patients who are in danger of death or need respiratory support. The model that has been provided can be utilized, in particular, as a tool that can assist physicians in predicting patients who are at risk and that can assist hospitals in effectively managing and organizing their resources.