Machine learning-enhanced echocardiography for screening coronary artery disease

Area of Science:

• Cardiology
• Artificial Intelligence
• Medical Imaging

Background:

• Coronary artery disease (CAD) screening is crucial for patient outcomes.
• Myocardial work (MW) and left atrial strain are established indicators for CAD.
• Novel approaches are needed to enhance CAD detection accuracy.

Purpose of the Study:

• To develop a machine learning-enhanced echocardiography approach for CAD screening.
• To leverage MW and left atrial strain features for improved CAD detection.
• To assess the diagnostic performance of the developed model.

Main Methods:

• A prospective study involving 818 patients undergoing coronary angiography.
• Extraction of novel echocardiography features using a machine learning algorithm.
• Development and validation of a gradient-boosting classifier using 59 echocardiographic features.

Main Results:

• The optimized model achieved an AUC of 0.852 (test group) and 0.834 (validation group).
• High sensitivity (0.952) indicates strong CAD detection capability.
• Low specificity (0.691) suggests a potential for increased false positives, with false-positive cases showing higher cardiac event susceptibility.

Conclusions:

• Machine learning-enhanced echocardiography effectively improves CAD detection using MW and left atrial strain.
• The developed model is valuable for estimating pre-test CAD probability.
• This approach shows clinical utility for screening CAD patients.