Leveraging Clinical Data for Early Heart Disease Prediction: A Machine Learning Approach With Interpretability
Emma Qumsiyeh1, Qassam Al-Wirdian1, Nur Sebnem Ersoz2
1Faculty of Engineering and Information Technology, Palestine Ahliya University, Bethlehem, Palestine.
Biomedical Engineering and Computational Biology
|May 19, 2026
Summary
Machine learning models, particularly Random Forest and K-Nearest Neighbors (KNN), show strong potential for accurate heart disease prediction. Explainable AI (SHAP) enhances model interpretability for clinical decision support.
Area of Science:
- Cardiology
- Medical Informatics
- Artificial Intelligence
Background:
- Heart disease is a major global cause of mortality.
- Early and accurate diagnosis is crucial for effective prevention and treatment.
- This necessitates advanced diagnostic tools for risk stratification.
Purpose of the Study:
- To develop and evaluate machine learning models for heart disease prediction.
- To compare the performance of Logistic Regression, Random Forest, KNN, and Decision Trees.
- To enhance model interpretability using SHAP values for clinical trust.
Main Methods:
- Utilized a publicly available clinical and demographic dataset.
- Performed data preprocessing including imputation, encoding, and normalization.
- Evaluated four classification algorithms (Logistic Regression, Random Forest, KNN, Decision Trees) using accuracy, precision, recall, and AUC-ROC metrics.
- Applied SHapley Additive exPlanations (SHAP) for model interpretability.
Main Results:
- Hyperparameter-optimized Random Forest and KNN models demonstrated superior predictive performance.
- SHAP analysis provided insights into feature importance and individual prediction explanations.
- The study confirmed the effectiveness of machine learning in predicting heart disease.
Conclusions:
- Interpretable machine learning models offer significant potential for early heart disease diagnosis and clinical decision support.
- SHAP enhances transparency and clinical trust in AI-driven diagnostic tools.
- Future work should focus on larger datasets and real-time applications to improve generalizability and clinical utility.
