Predicting COVID-19 Models for Death with Three Different Decision Algorithms: Analysis of 600 Hospitalized Patients

COVID-19 virus has caused the biggest pandemic in a decade. The acute respiratory syndrome caused by this virus can lead to the death of patients. Death is very likely in people with severe forms of lung disease. Early identification of patients with severe disease can be very effective in the prevention of death outcomes with improves triage strategies and timely medical actions. The aim of this study was the prediction of COVID-19 models for death with three different decision algorithms with analysis of hospitalized patients. In this study, in a retrospective analysis of 600 COVID-19 patients, we apply three decision tree algorithms including the C5.0, CRT, and CHAID using all related factors to the disease including demographic data, history of exposure, clinical signs, and symptoms, laboratory results, chest X-ray or computed tomography (CT) scans, underlying illness, treatment steps, and outcomes of each patient to build several models predicting the death of Covid-19 infection. The accuracy of the models was above 90%. Overall, in our retrospective analysis, age, hypertension, lung disease, O2Sat, diabetes, and body temperature, respectively are the most important factors that can affect the mortality rate of COVID-19 patients. Among them, age, and hypertension are common in our applied three models. The design of such models and apply in hospitals can help to improve disease management and decrease the mortality rate spatially in about recent pandemic.