Predicting COVID-19 severity using machine learning: the role of clinical parameters, blood groups, and vaccination status across age groups.

OBJECTIVES: COVID-19 has had strong impacts on both global health systems and economies. Therefore, understanding disease severity and progression is crucial. This cross-sectional study aims to investigate possible link between clinical factors, laboratory parameters, and age on COVID-19 severity. Machine-learning (ML) methods were used to analyze a dataset to explore the relationships between these variables and disease progression. METHODS: In this study, 177 hospitalized COVID-19 patients were divided into five age groups. A range of biomarkers, including hematological parameters, electrolytes, and liver and renal function tests, were analyzed. Several parameters had acceptable sensitivity and specificity and indicated cut-off levels between patient age groups. Furthermore, ML was used to find an interaction between blood group, pneumococcal vaccine status, and COVID-19 symptoms to explore potential relationships. RESULTS: The main findings were that various biomarkers were differentially regulated in COVID-19 patients across different age groups. Furthermore, the interpretation of the obtained results indicated a significant association between the presence of symptoms during the course of the illness and pneumococcal vaccination status. Additionally, feature selection analysis employing machine learning techniques, specifically Random Forest and Mutual Information, identified the most important features within the analyzed dataset. CONCLUSIONS: Although blood group and Pneumococcal vaccine showed weak influence on the presence of COVID-19 symptoms, our findings shed light on the relative importance of the most influencing predictors on COVID-19 severity, which may provide valuable-insights for better risk assessment.