Predictive Modelling of COVID-19 Severity using Machine Learning Algorithms

Authors

Naralay Viji Paul, Amit Shah

Abstract

The COVID-19 pandemic has placed unprecedented stress on global healthcare systems, amplifying the need for predictive tools to manage patient care more effectively. This study addresses the challenge of identifying patients at high risk of severe COVID-19 outcomes by developing machine learning models that leverage patient demographic data, clinical conditions, and pre-existing health issues. The dataset, obtained from the Mexican government's open-source COVID-19 patient records, includes comprehensive information on key comorbidities, hospitalization status, and patient demographics. Through a rigorous data preprocessing pipeline that includes cleaning missing values, encoding categorical variables, and scaling features, the dataset is prepared for advanced machine learning analysis. Various algorithms, including Logistic Regression, Random Forest, and Support Vector Machines, are applied to build predictive models. Model performance is measured using metrics such as accuracy, precision, recall, and the Area Under the Receiver Operating Characteristic Curve (AUC-ROC), with cross-validation techniques ensuring model generalizability. This research has the potential to provide actionable insights for healthcare professionals by enabling more efficient triaging of patients, improving resource allocation such as ICU beds and ventilators, and potentially reducing the burden on overwhelmed healthcare infrastructures during pandemics.

Keywords

Include at least 5 keywords or phrases

Predictive Modelling of COVID-19 Severity using Machine Learning Algorithms. Naralay Viji Paul, Amit Shah. 2025. IJIRCT, Volume 11, Issue 2. Pages 1-14. https://www.ijirct.org/viewPaper.php?paperId=2503119