SARS-COV-2 MORTALITY RISK PREDICTION USING MACHINE LEARNING TECHNIQUE

Abstract

Coronavirus referred to as COVID-19 has had adverse effects in every possible aspect such as loss of economy, infrastructure, and moreover human life. In the era of growing technology: Artificial intelligence and machine learning can help find a way in reducing mortality, and in the same regard, we have created a prediction model for mortality of in-hospital COVID-19 patients. We used the dataset of 146 countries which consists of laboratory samples of around 2,670,000 confirmed COVID-19 cases. This study presents a Machine Learning model which will assist hospitals and medical facilities in determining who requires immediate attention and who must be given priority for hospitalization when the system is overburdened, or the facility is filled with patients who are not that severe and eliminate any delays in providing needed care to extremely severe patients first. As a result, the overall accuracy of the mortality rate prediction demonstrated is 91.26%. We evaluated different machine learning algorithms namely decision tree (DT), support vector machine (SVM), random forest (RF), logistic regression (LR), and k-nearest neighbor (K-NN) for mortality risk prediction COVID 19 affected patients admitted in hospitals. This proposed research study sheds light upon the identification of most relevant features and concerning symptoms. To perform an indepth examination and assess the results of classifiers, we used different performance measures on the developed model.