Fuzzy Rule-Based Classification System for Assessing Coronary Artery Disease.

Area of Science:

• Cardiology
• Artificial Intelligence
• Medical Informatics

Background:

• Coronary artery disease (CAD) diagnosis often relies on invasive procedures.
• Noninvasive methods for predicting CAD severity are crucial for patient management.
• Myocardial perfusion scans (MPS) provide valuable functional information about the heart.

Purpose of the Study:

• To evaluate the accuracy of a fuzzy rule-based classification system for noninvasively predicting CAD.
• To assess the system's ability to classify CAD severity based on MPS and clinical data.
• To develop an accurate, noninvasive tool for CAD risk stratification.

Main Methods:

• A cross-sectional study involving 115 patients with collected clinical, MPS, and coronary angiography data.
• Fuzzy rule-based classification using membership functions for medical variables and adjusted certainty factors (CF j).
• Implementation of a system with 144 rules, refined by deleting redundant rules to optimize classification.

Main Results:

• The fuzzy classification system achieved 92.8% accuracy when results were expert-selected and 91.9% when equation-derived.
• The system successfully classified patients into four categories: normal, single-vessel, double-vessel, and triple-vessel stenosis.
• The developed fuzzy logic system demonstrated high predictive accuracy for coronary artery disease.

Conclusions:

• Fuzzy rule-based classification offers a highly accurate noninvasive method for predicting coronary artery disease.
• This approach can effectively stratify patients based on CAD severity using myocardial perfusion scans and clinical data.
• The study highlights the potential of AI in improving noninvasive diagnostic capabilities for cardiovascular diseases.