AN ENHANCED HYBRID PCA–IDFOA FEATURE OPTIMIZATION MODEL FOR ACCURATE CARDIOVASCULAR DISEASE PREDICTION USING RANDOM FOREST
: https://doi.org/10.65725/JCISE/2/4/003
JOURNAL OF COMPUTATIONAL INTELLIGENCE SCIENCE AND ENGINEERING (JCISE) 
ISSN: 3107-8168
Volume 2 Issue 4, Oct – Dec 2026

Abstract: 

Cardiovascular disease (CVD) remains a major cause of mortality and requires accurate early-risk prediction. Clinical datasets commonly contain correlated, redundant, noisy and differently scaled attributes that can affect machine-learning performance. This paper proposes an enhanced hybrid Principal Component Analysis–Improved Dragonfly Optimization Algorithm (PCA–IDFOA) feature optimization model with Random Forest for CVD prediction. The framework first performs data cleaning, missing-value treatment, categorical encoding and Z-score normalization. PCA is then applied to transform the original clinical feature space into a compact representation while reducing multicollinearity and preserving important variance. IDFOA subsequently searches the reduced feature space to identify an informative subset by balancing predictive performance and feature compactness. The optimized features are supplied to a Random Forest classifier, with optional hyperparameter optimization and explainability as enhanced stages. The model is evaluated using accuracy, precision, recall, specificity, F1-score and ROC-AUC. The source study reports 0.89 accuracy, precision, recall and F1-score for its reported optimized Random Forest model. The present paper retains those reported values as baseline evidence and proposes a more detailed experimental methodology for validation on a new database. The framework aims to reduce feature redundancy, improve computational efficiency and support reliable clinical decision support for cardiovascular-risk assessment.

Authors:  P. Kavitha, Dr. R. Manicka Chezian

Keywords: Cardiovascular Disease; Principal Component Analysis; PCA; Improved Dragonfly Optimization Algorithm; IDFOA; Feature Optimization; Random Forest; Machine Learning; Clinical Decision Support.