To classify prevalent hypertension status in the general population using explainable machine learning models.
Approach:
Model Evaluation: Models were compared based on classification accuracy and AUC values, with additional assessments including calibration and decision curve analysis to evaluate clinical applicability.
Key Findings:
32.5% of the participants were classified as hypertensive out of 4,606 total participants.
The eXtreme Gradient Boosting (XGBoost) model achieved the highest performance with an AUC of 0.8461.
Top features influencing hypertension classification included age, smoking, urine microalbumin, educational level, diabetes, BMI, sex, triglyceride, income level, and family history of hypertension.
Interpretation:
The study developed a machine learning model that effectively identifies prevalent hypertension cases, enhancing screening capabilities.
Limitations:
The study is limited to a specific geographic area, which may affect generalizability.
The reliance on self-reported data may introduce bias.
Conclusion:
The developed machine learning model demonstrates superior performance in identifying prevalent hypertension, facilitating rapid population screening.
