Predicting early-stage coronary artery disease using machine learning and routine clinical biomarkers improved by...

Area of Science:

• Cardiology
• Medical Informatics
• Artificial Intelligence

Background:

• Coronary artery disease (CAD) is a leading cause of mortality with modifiable risk factors.
• Accurate pre-test probability assessment is vital for diagnosing obstructive CAD, but current models face accuracy challenges.
• Machine learning (ML) offers potential for early CAD detection and improved patient outcomes.

Purpose of the Study:

• To identify early-stage coronary artery disease (CAD) using machine learning (ML) models.
• To evaluate the efficacy of ML in conjunction with clinical and laboratory tests for CAD detection.
• To improve the performance of ML models through the generation of high-quality virtual patient data.

Main Methods:

• Utilized data from 3316 patients in the Ludwigshafen Risk and Cardiovascular Health (LURIC) study.
• Developed an ML pipeline incorporating a comprehensive set of clinical and laboratory attributes.
• Employed five methods for generating virtual patient data to augment the training datasets.
• Validated findings using data from the Young Finns Study (YFS) for generalizability.

Main Results:

• The application of virtual augmented data improved ML model performance, increasing accuracy by approximately 5%.
• Random Forests (RFs) and Gradient Boosting (GB) models showed enhanced accuracy, sensitivity, and specificity after data augmentation.
• RF accuracy increased from 0.75 to 0.79, and GB accuracy from 0.76 to 0.80.
• Significant improvements in sensitivity and specificity were observed for both RFs and GB models with augmented data.

Conclusions:

• Routine laboratory markers, age, sex, and smoking status can accurately predict angiographic CAD.
• ML models utilizing these factors show potential to reduce the need for invasive diagnostic procedures.
• The findings were validated in a younger population via the YFS, confirming applicability to atherosclerotic vascular disease.