A risk prediction model based on machine learning for early cognitive impairment in hypertension: Development and...

Area of Science:

• Neurology
• Cardiology
• Data Science

Background:

• Clinical guidelines advocate for early cognitive impairment detection in hypertensive individuals using risk prediction tools.
• Current risk prediction tools rely on various risk factors for early identification.

Purpose of the Study:

• To develop a superior machine learning model for predicting early cognitive impairment risk in hypertensive patients.
• To utilize easily collected variables for optimized risk assessment strategies.

Main Methods:

• A cross-sectional study involving 733 hypertensive patients in China.
• Development and comparison of three machine learning models: logistic regression (LR), XGBoost (XGB), and gaussian naive bayes (GNB).
• Model performance evaluated using AUC, accuracy, sensitivity, specificity, and F1 score; feature importance assessed via SHAP analysis.

Main Results:

• XGBoost model demonstrated superior performance with an AUC of 0.88, accuracy of 0.81, sensitivity of 0.84, specificity of 0.80, and F1 score of 0.59.
• Significant predictors identified: hip circumference, age, education level, and physical activity.
• The XGBoost model outperformed LR and GNB classifiers in predictive accuracy.

Conclusions:

• The XGBoost model, utilizing hip circumference, age, education, and physical activity, shows excellent predictive performance.
• This model holds promise for enhancing cognitive impairment risk prediction in hypertensive patients within clinical practice.