An Interpretable Ensemble Framework Towards Efficient Hypertension Risk Assessment

Hypertension, a worldwide health issue, increases the risk of serious cardiovascular illnesses, whether untreated or undetected. The proposed prediction framework uses classical machine learning algorithms and Explainable AI (XAI)-enhanced ensemble learning to improve the diagnosis and understanding of high blood pressure risk factors. The dataset is available on the Kaggle titled "Diabetes, Hypertension, and Stroke Prediction repository", and consists of 26,083 samples with 14 clinical factors. Hypertension was predicted using traditional machine learning models such as Decision Tree, Random Forest, Support Vector Machine, Gradient Boosting, and AdaBoost. Besides, an ensemble approach: voting classifier is introduced, combining the individual models to enhance performance. Experimental data indicate that the Voting Classifier, Random Forest, and Decision Tree obtained a perfect accuracy of 100%, topping other models, while Gradient Boosting and Support Vector Machine achieved 97.6% and 99.9% accuracy, respectively. The Area Under the Curve scores further confirmed the robustness of the models. To ensure the model’s transparency, Shapley Additive Explanations (SHAP) plots were applied, highlighting significant factors impacting hypertension prediction. The SHAP summary and decision plot added greater insights into the decision-making process, assuring dependability in healthcare applications. The results conclude that machine learning and ensemble approaches can successfully detect hypertension with high accuracy, while XAI techniques provide the model’s interpretability. Additionally, the proposed framework highlights the potential of AI-driven diagnostics in assisting early hypertension identification, which may lead to better disease prediction and faster medical intervention.