Web-based cardiovascular disease risk prediction using machine learning.

Area of Science:

• Biomedical Informatics
• Machine Learning in Healthcare
• Cardiovascular Disease Research

Background:

• Cardiovascular disease (CVD) is a leading global cause of mortality, necessitating advanced risk prediction tools.
• Machine learning (ML) offers potential for enhanced medical decision-making through complex healthcare data analysis.
• Effective ML model performance hinges on the relevance and quality of input features.

Purpose of the Study:

• To compare four feature-selection strategies for identifying optimal predictors of CVD risk.
• To evaluate the predictive performance of ML models developed using distinct feature sets.
• To enhance the interpretability of ML models for clinical application.

Main Methods:

• Utilized National Health and Nutrition Examination Survey (NHANES) data (2021-2023).
• Compared Pearson correlation + Chi-squared, ADT-based scoring, CVFE, and HFE for feature selection.
• Developed and assessed Random Forest (RF), SVM, and XGBoost models.
• Employed SHapley Additive exPlanations (SHAP) for model interpretability.

Main Results:

• The HFE approach combined with SVM achieved the highest accuracy (82.84%) and AUC (0.9027).
• Identified key predictors: age, total cholesterol, hypertension history, cholesterol medication use, recent prescription use, smoking history, income-to-poverty ratio, gender, education, and red cell distribution width.
• A web application was developed for CVD risk prediction using the HFE-selected features.

Conclusions:

• Strategic feature selection significantly improves the accuracy and interpretability of CVD risk prediction models.
• The HFE method offers a robust approach for identifying critical CVD risk factors.
• This data-driven strategy can assist clinicians in cardiovascular risk assessment and preventive care planning.