Covid-19 risk factors prediction using machine learning: South Sudan and Eastern Democratic Republic of the Congo dataset

Abstract

The COVID-19 pandemic has posed significant challenges to global health systems, particularly in resource-constrained regions such as South Sudan and the Eastern Democratic Republic of the Congo (DRC). Understanding the risk factors associated with COVID-19 mortality is crucial for effective prevention and management. In this research paper, we leverage machine learning techniques to analyze a dataset collected from these regions, encompassing sociodemographic characteristics, COVID-19 exposures, symptoms, health history, and laboratory test results. By employing advanced statistical models and data mining algorithms, we identify significant risk factors associated with COVID-19 mortality. The machine learning models that had been used in this study are Logistic Regression, Decision Tree, Random Forest and XGBoost. Based on machine learning models run, the highest performing machine learning model for mortality risk prediction on Covid-19 positive patient in South Sudan and Eastern Democratic Republic of the Congo is Random Forest model with 87.01% accuracy rate, 86.67% precision, 86.67% recall rate, and F1-Score of 86.67%. The SHAP model employed show that the appearance of the Covid-19 positive patient at enrolment, the patient nationality, and presence of any symptoms related to Covid-19 infection is the significant contributor toward mortality condition.