Machine Learning for Hypertension Prediction in U.S. University-Aged Students

Insights from NIH All of Us Data

Abstract

Hypertension is a major risk factor for cardiovascular disease and early detection is essential for preventing long-term complications. In this work, we investigate AI-driven hypertension prediction using the National Institutes of Health (NIH) All of Us research dataset, focusing on university-aged individuals in the United States. We develop a machine learning–based detection framework utilizing five feature categories: demographics, clinical laboratory tests, vital health measurements, family medical history and lifestyle and behavioral factors. Multiple supervised learning models are evaluated, including Decision Tree, K-Nearest Neighbor, Support Vector Machine and Extreme Gradient Boosting (XGBoost). XGBoost achieved the best performance, obtaining an accuracy of 84.88% and sensitivity of 0.787, outperforming all baseline classifiers. These results establish a strong baseline for hypertension risk modeling in university populations.