Early Detection Of Hypertension Using Stacked Ensemble Learning with SMOTE And Feature Selection

Abstract

Hypertension (HTN), or high blood pressure (BP), is a serious health condition that arises when BP levels remain consistently above normal. It is often linked to modern lifestyle changes and lack of regular physical activity. Early detection of HTN is crucial, as it enables timely treatment and can help prevent life-threatening complications. In this study, we propose an improved machine learning (ML)-based approach for early detection of HTN. Our method uses stacking ensemble techniques, both with and without hyper parameter tuning (HPT), and applies the Synthetic Minority Over-sampling Technique (SMOTE) to handle class imbalance. We also perform feature selection (FS) using correlation scores to reduce over fitting and improve model performance. The models are evaluated using various metrics, and experimental results show that our stacking approach achieves a high accuracy of 99%, significantly outperforming previous models. This system can help patients quickly assess their HTN risk without waiting for a medical expert.