Risk factor identification mechanism for coronary artery disease based on multiple cross-filtering and binary cuckoo...

Area of Science:

• Cardiology
• Biomedical Informatics
• Machine Learning

Background:

• Coronary artery disease (CAD) poses a significant public health challenge.
• Current risk factor identification methods exhibit feature bias and limited multidimensional analysis.
• Reliable CAD risk assessment requires improved feature selection techniques.

Purpose of the Study:

• To develop a novel two-stage mechanism for accurate CAD risk factor identification.
• To overcome limitations of existing methods, including feature preference bias and insufficient evaluation.
• To enhance the reliability and predictive performance of CAD risk assessment tools.

Main Methods:

• A two-stage mechanism integrating multiple cross-filtering and binary cuckoo search (BCS).
• Stage 1: Feature evaluation using chi-square test (relevance), mutual information (information richness), and Fisher score (distance).
• Stage 2: Random forest classifier optimizing feature selection via BCS with classification accuracy as the objective function.

Main Results:

• The proposed method achieved 89% accuracy, 0.87 precision, 0.91 recall, 0.89 F1-score, and 0.93 AUC on the UCI CAD dataset.
• Performance metrics significantly outperformed existing homogeneous models.
• Demonstrated superior computational efficiency and predictive capabilities.

Conclusions:

• The novel two-stage mechanism offers a valuable tool for CAD risk assessment.
• The method effectively reduces bias and enhances multidimensional evaluation of risk factors.
• Results indicate significant improvements in predictive performance and computational efficiency compared to existing models.