Machine Learning–Based Hypertension Risk Stratification Using Logistic Regression and XGBoost

Hypertension is a critical global health concern, necessitating accurate prediction models and effective prescription decisions to mitigate its risks. This study proposes a hybrid machine learning approach using Logistic Regression and XGBoost to predict hypertension and recommend optimal prescriptions based on a clinical dataset of 333 patients. The model leverages patient demographics such as age, gender, systolic pressure and diastolic pressure as input features, along with diagnostic variables, to classify hypertension risk and suggest appropriate antihypertensive medications. Model performance was assessed by examining the area under the receiver operating characteristic curve (AUC), accuracy, precision, recall, and F1-score. Furthermore, the analysis of learning rate, loss value, and confusion matrix further confirm the model's predictive accuracy. The XGBoost model outperformed the logistic regressor, achieving accurate and precise hypertension prediction with an F1-score equal to 1.