Hypertension Risk Prediction Using Optimized and Interpretable Learning Models

Hypertension (HTN) or high blood pressure, is a widespread global problem and is a major risk factor for heart disease, premature death and long term inability. Early detection of individuals is therefore very important for people who may be at risk. This study emphasizes the role of preventive strategies by proposing a machine learning (ML) based framework for accurate hypertension risk prediction. Leveraging recent advances in computational intelligence, multiple ML models including XGB, RF, Logistic Regression (LR), KNN, and SVC are developed and evaluated. To enhance predictive performance, an optimal feature selection strategy based on Layer-wise Annealing Optimization (LAO) is employed. The experimental evaluation, conducted with 10-fold cross validation, demonstrates that the LR model combined with LAO selected features shows the maximum accuracy 99.44%. Furthermore, XAI techniques, SHAP and LIME are applied to explain the models predictions and identify the most important risk factors that influence hypertension. The mentioned system provides a transparent, accurate, and reliable system which can assist healthcare professionals in early risk assessment and intervention, thereby addressing critical challenges in modern healthcare.