Toward Reliable Coronary Heart Disease Prediction: Integrating Multi-source Data with Ensemble Machine Learning.

Area of Science:

• Cardiology
• Medical Informatics
• Machine Learning

Background:

• Coronary heart disease (CHD) is a leading global cause of mortality.
• Early detection and risk assessment are crucial for patient management and reducing fatality rates.
• Machine learning (ML) offers a powerful tool for analyzing clinical data to predict heart disease.

Purpose of the Study:

• To propose a reliable ML model for predicting coronary heart disease.
• To integrate multi-source heart disease data with various ML algorithms.
• To enhance the accuracy and robustness of CHD prediction.

Main Methods:

• Utilized four public heart disease datasets (Cleveland, Hungary, Switzerland, VA Long Beach).
• Applied diverse ML models: logistic regression, Naive Bayes, random forest, XGBoost, KNN, decision trees, SVM.
• Implemented an ensemble-learning approach, combining high-performing models.
• Addressed class imbalance using synthetic minority oversampling technique (SMOTE).

Main Results:

• The proposed ensemble model achieved 98.46% accuracy.
• Achieved 96% precision, 100% recall, and 98% F1-score.
• Demonstrated superior performance compared to individual models.

Conclusions:

• The developed ensemble ML model is effective and robust for coronary heart disease prediction.
• The approach facilitates timely intervention and personalized treatment strategies.
• Highlights the potential of integrated ML models in cardiovascular risk assessment.