부산대학교 도서관

Predicting the first signs of heart disease is essential for timely diagnosis and treatment, as this condition is a significant cause of death globally. Knowing in advance who is at risk for developing heart disease allows doctors to provide preventative care and treatment options specifically suited to each patient. By combining Correlation-based Feature Selection (CFS), Grey Wolf Optimization Algorithm (GWOA), and seven other conventional machine learning classification methods, we develop a novel hybrid model that improves the accuracy of heart disease prediction. The hybrid model aims to enhance the predictivity of machine learning classifiers through optimal feature selection. Experiments in this research achieve an impressive 99.01% accuracy, 99.10% precision, 99.55% sensitivity, 97.53% specificity, and 99.32% F-measures, demonstrating the approach's potential to aid medical professionals in early diagnosis and risk assessment.