Prediction of Cognitive Impairment in Elderly Hypertensive Patients in the United States Using Machine Learning Algorithms: A Cross-Sectional Study

Background: This study aimed to evaluate the utility of machine learning (ML) algorithms in predicting cognitive impairment among elderly individuals with hypertension in the United States and to identify key associated risk factors. Methods: Data were obtained from 19,931 participants enrolled in the 2011–2012 and 2013–2014 cycles of the National Health and Nutrition Examination Survey (NHANES). The dataset was randomly split into training and test sets (70:30). Seven ML algorithms—logistic regression (LR), extreme gradient boosting (XGB), decision tree (DT), categorical boosting (CatBoost), random forest (RF), light gradient boosting machine (LGBM), and support vector machine (SVM)—were trained to predict cognitive impairment. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), calibration curves, and decision curve analysis (DCA). The SHapley Additive exPlanation (SHAP) method was applied for feature importance interpretation. Results: Among all models, LR exhibited the best overall performance, achieving an AUC of 0.791 on the test set, with superior F1-score and accuracy. Calibration plots demonstrated good agreement between predicted and observed outcomes. DCA confirmed the clinical utility of the LR model. SHAP analysis identified the key variables contributing to model predictions. A web-based calculator based on the final LR model, incorporating 12 predictors, is available at: https://cognitiveimpairment.shinyapps.io/cognitiveimpairment. Conclusion: An interpretable ML model was developed and validated to predict the risk of cognitive impairment in elderly hypertensive patients in the United States. This model enables clinicians to quickly identify high-risk patients, which in turn supports more effective prevention and intervention strategies.