Non-invasive Coronary Artery Disease Screening Based on Electrocardiogram Characteristics and Clinical Risk Factors

Area of Science:

• Cardiology
• Biomedical Engineering
• Machine Learning

Background:

• Coronary artery disease (CAD) is a major global health concern, with current diagnostic methods like coronary angiography being invasive and costly.
• There is a need for accessible, non-invasive screening tools for early CAD detection.

Purpose of the Study:

• To develop an automated, non-invasive model for detecting CAD using electrocardiogram (ECG) and clinical data.
• To achieve diagnostic performance comparable to current clinical standards.

Main Methods:

• Extracted morphological ECG features and heart rate variability (HRV) from short-duration Lead-II ECG recordings.
• Integrated clinical information with ECG-derived features.
• Trained and evaluated machine learning classifiers including SVM, RF, KNN, GNB, and MLP.

Main Results:

• Classifiers integrating both clinical and ECG features outperformed those using single feature sets.
• The Random Forest (RF) classifier achieved the highest performance with 94% average testing accuracy and 92% G-mean.
• Achieved 95(± 0.04)% accuracy with 5-fold cross-validation.

Conclusions:

• The proposed automated CAD detection model, utilizing brief ECG recordings and clinical data, demonstrates high accuracy and explainability.
• This non-invasive approach is suitable for integration into wearable technology and telemonitoring systems for widespread, early CAD diagnosis.