AI-CADR: Artificial Intelligence Based Risk Stratification of Coronary Artery Disease Using Novel Non-Invasive...

Area of Science:

• Cardiology
• Biomarkers
• Machine Learning

Background:

• Coronary artery disease (CAD) is a leading cause of mortality and sudden cardiac arrest globally.
• Timely diagnosis and risk stratification are crucial for managing CAD and improving patient survival.
• Current risk assessment often relies on invasive procedures, highlighting the need for non-invasive methods.

Purpose of the Study:

• To develop and validate a novel methodological framework for non-invasive risk stratification of CAD.
• To explore the utility of novel clinical, chemical, and molecular cardiac biomarkers for CAD risk assessment.
• To integrate machine learning techniques with biomarkers for enhanced early detection of CAD.

Main Methods:

• Utilized a specially collected dataset of novel cardiac biomarkers (clinical, chemical, molecular).
• Employed K-fold cross-validation for optimizing machine learning classifier and regressor parameters.
• Applied ten machine learning classifiers for classification tasks and eleven regression approaches for regression tasks.

Main Results:

• Classification tasks achieved high accuracy: 82.58% for affected vessels, 86.26% for Gensini group, and 90.91% for CAD severity.
• Regression tasks showed moderate performance: R-squared values of 0.58 for stenosis percentage and 0.56 for Gensini score.
• Identified optimal biomarker and machine learning model combinations for the proposed risk stratification framework.

Conclusions:

• The proposed framework offers a robust, non-invasive approach to CAD risk stratification by combining biomarkers and machine learning.
• This novel 'biomarkers-ML combination' approach facilitates early detection and has the potential to significantly impact patient management.
• The framework demonstrates a significant advancement over existing state-of-the-art methods in CAD risk assessment.