A Novel Hybrid Feature Selection Framework with Dimensionality Reduction for Early Cardiovascular Disease Detection

Abstract

Cardiovascular disease (CVD) is one of the leading causes of mortality worldwide, necessitating the development of accurate and efficient predictive models for early diagnosis and intervention. However, the high dimensionality, redundancy, and heterogeneity of clinical data often limit the performance of conventional machine learning (ML) approaches. To address these challenges, this study proposes a hybrid feature selection with dimensionality reduction framework for CVD detection (HFSDR-CVD). The proposed framework integrates principal component analysis for dimensionality reduction, mutual information-based feature ranking for relevance assessment, and recursive feature elimination for optimal feature subset selection. The optimized feature space is subsequently used to train and evaluate multiple ML classifiers, including logistic regression, support vector machine, random forest, gradient boosting, and extreme gradient boosting (XGBoost). Experimental evaluation demonstrates that the proposed framework effectively reduces the original feature set from 13 to 6, achieving a feature reduction rate of 53.85% while preserving critical diagnostic information. Among the evaluated classifiers, XGBoost achieved the highest performance with an accuracy of 98.31%, precision of 98.06%, recall of 97.95%, F1-score of 98.00%, and ROC-AUC of 99.02%. Comparative analysis with recent state-of-the-art CVD prediction methods confirms the superiority of the proposed HFSDR-CVD framework in terms of predictive accuracy, feature optimization, and computational efficiency.