Early identification of CVD in hypertensive patients remains a clinical priority. A study published in Archives of Cardiovascular Diseases developed and validated a machine learning–based prediction model designed to improve early CVD screening and individualized risk assessment.

The study analyzed data from 2,781 hypertensive participants in the National Health and Nutrition Examination Survey (2009–2018). Using a combination of Least Absolute Shrinkage and Selection Operator (LASSO), Boruta, and Recursive Feature Elimination techniques, eight crucial predictors were selected: neutrophil-lymphocyte ratio, waist-to-height ratio, age, high-density and low-density lipoprotein cholesterol, kidney disease, sleep disturbance, and diabetes.

Among four machine learning algorithms tested, the Balanced Bagging Classifier achieved the best overall performance after 10-fold cross-validation and independent test validation. SHapley Additive exPlanations (SHAP) analysis further confirmed that inflammatory and metabolic markers such as neutrophil-lymphocyte ratio and waist-to-height ratio were the most influential factors.

The findings demonstrate that interpretable machine learning models can improve cardiovascular risk prediction in hypertensive patients. Broader validation could establish this model as a useful clinical tool for targeted prevention and decision support in hypertension management.