Integration of Different Risk Assessment Tools to Improve Stratification of Patients with Coronary Artery Disease

Area of Science:

• Cardiology
• Biomedical Informatics
• Machine Learning

Background:

• Cardiovascular disease (CVD) poses significant healthcare burdens, particularly in aging European populations.
• Clinical guidelines advocate for risk assessment tools to predict CVD events, but existing methods have limitations.
• Accurate prediction of cardiovascular events like hospitalization or death is crucial for patient management.

Purpose of the Study:

• To develop and validate improved methodologies for cardiovascular disease risk assessment.
• To address drawbacks of current risk prediction tools through innovative computational approaches.
• To enhance the accuracy, interpretability, and flexibility of CVD risk prediction models.

Main Methods:

• A novel approach combining multiple risk assessment tools using naïve Bayes classifiers and genetic optimization.
• Personalized risk assessment achieved through subtractive clustering in a reduced-dimensional space.
• Validation conducted on two Acute Coronary Syndrome with Non-ST-Elevation Myocardial Infarction (ACS-NSTEMI) patient datasets.

Main Results:

• The proposed methodologies demonstrated improved performance over existing CVD risk assessment tools.
• Achieved maximum sensitivity of 79.8%, specificity of 83.8%, and geometric mean of 80.9%.
• Ensured clinical interpretability, adaptability for new risk factors, and better handling of missing data.

Conclusions:

• The developed methods offer a more robust and adaptable approach to cardiovascular disease risk prediction.
• These advancements provide a valuable alternative to standard, single CVD risk assessment tools in clinical practice.
• The enhanced capabilities address key limitations, paving the way for more personalized and effective patient care.