Assessing the Comparative Effectiveness of Ensemble CatBoost Versus XGBoost Models in Predicting Heart Disease Outcomes

Abstract

Cardiovascular Diseases (CVDs) are a common form of heart disease that remains a significant global cause of mortality, responsible for more than 30% of all fatalities. Without intervention, the global fatality count is projected to reach 22 million by 2030. Arterial plaque buildup may impede blood flow, potentially causing heart attacks or strokes. A combination of risk causes, including lack of physical exercise, a poor diet, and the excessive use of alcohol and tobacco, mostly causes heart disease. Precise classification of cardiovascular disease is crucial for cardiologists to provide suitable treatment to patients. Diagnosis and prognosis are critical medical concepts in this regard. Machine learning has become more prevalent in the medical domain. Utilizing machine learning for the classification of cardiovascular disease incidence may aid diagnosticians in minimizing misdiagnosis. By using CatBoost and XGBoost models, it is possible to effectively predict cardiovascular illnesses. We use performance assessment criteria, such as precision, recall, F1score, and accuracy assessments, to conduct a comprehensive analysis of our approaches. XGBoost achieved an accuracy rate of 91.33, a precision of 88. 38, a recall of 88.63, and an F1 value of 89.75. However, CatBoost achieved an accuracy rate of 94. 09, a precision of 91.38, a recall of 89.83%, and an F1 value of 90.38. CatBoost is identified as the most effective ensemble method. This heart disease prediction model may serve as an adjunctive diagnostic tool for physicians, providing accurate and rapid predictions.