Machine learning framework for atherosclerotic cardiovascular disease risk assessment

Area of Science:

• Cardiovascular Disease Research
• Machine Learning in Healthcare
• Biomedical Data Science

Background:

• Atherosclerotic cardiovascular disease (ASCVD) is a leading global cause of mortality.
• Identifying individual risk factors is crucial for effective prevention strategies.

Purpose of the Study:

• To identify individual risk factors for ASCVD using machine learning (ML) approaches.
• To develop and validate ML models for predicting ASCVD risk.

Main Methods:

• A cohort-based cross-sectional study of 500 ASCVD participants was conducted.
• Multiple ML models including Naive Bayes, SVM, RT, KNN, ANN, GAM, and LR were employed.
• Model performance was evaluated using metrics like accuracy, sensitivity, specificity, PPV, NPV, LR+, LR-, and AUC.

Main Results:

• Models demonstrated high accuracy, ranging from 95.7% to 98.1%.
• The Artificial Neural Network (ANN) model showed superior performance with 98.1% accuracy, 99.1% specificity, and 99.4% AUC.
• Key predictors identified by ANN included female sex, age, smoking, and metabolic syndrome.

Conclusions:

• ANN emerged as the optimal model for predicting ASCVD risk.
• The identified risk factors (female sex, age, smoking, metabolic syndrome) can inform targeted prevention planning.