Cardiovascular Disease Risk Classification Using Machine Learning with Weighted Feature Fusion and Explainable AI on Bangladeshi Clinical and Lifestyle Data

Asif, Tasnimul Intazam (International University of Business Agriculture and Technology (IUBAT))
Ray, Bishwaprotap (International University of Business Agriculture and Technology (IUBAT))
Hossain, Md. Alomgir (International University of Business Agriculture and Technology (IUBAT))
Imran, Faisal (Department of Computer Science and Engineering (CSE) & CSIT, Shanto-Mariam University of Creative Technology, Dhaka, Bangladesh)
Barua, Prime Biswajit (Department of Computer Science and Engineering, International University of Business and Agricultural Technology)
Anisha, Nishat Salsabil (International University of Business Agriculture and Technology (IUBAT))
Minhaj, Ariful Haque (International University of Business Agriculture and Technology (IUBAT))
Roy, Amit (International University of Business Agriculture and Technology (IUBAT))

Indonesian Journal of Electrical Engineering and Informatics (IJEEI)
Vol 14, No 1: March 2026

Cardiovascular disease (CVD) is one of the top causes of death across the world, and there is a need to develop early risk prediction models that can be accurate and interpreted. This study introduces a weighted feature fusion (WFF) model of machine learning to integrate clinical, lifestyle, and engineered features into an integrated machine learning model to improve the classification of CVD risk and the interpretability of the model. Several classifiers, such as the Random Forest (RF), K-Nearest Neighbors (KNN), XGBoost (XGB), Bagging, Decision Tree were trained and tested based on fusion-based methods. The experimental findings indicated that the highest classification accuracy of the model at 91% obtained by the Random Forest model. Moreover, the model was better in terms of discrimination as ROCAUC scores were over 0.980447in all categories of CVD risk. Explainable AI algorithms, such as SHAP and LIME were used to provide transparency, when combined with feature fusion, leads to a significant improvement in accuracy, reliability, and interpretability of CVD risk prediction models that can lead to the development of data-driven healthcare decision support systems of trust.