Risk stratification of coronary artery bypass patients using an artificial intelligence electrocardiogram-derived age

Area of Science:

• Cardiology
• Artificial Intelligence
• Medical Prognostics

Background:

• Coronary artery bypass grafting (CABG) outcomes can be influenced by patient comorbidities and physiological status.
• Traditional risk stratification may not fully capture individual patient risk profiles.

Purpose of the Study:

• To evaluate the prognostic significance of artificial intelligence electrocardiogram-derived age (AI ECG-age) in predicting outcomes after isolated CABG.
• To determine if AI ECG-derived age, beyond chronological age, offers additional predictive value.

Main Methods:

• Utilized preoperative electrocardiograms from 13,808 isolated CABG patients.
• Calculated AI ECG-age using convolutional neural networks and determined the age gap (AI ECG-age minus chronological age).
• Analyzed associations between age gap, comorbidities, operative outcomes, and long-term survival using multivariable regression.

Main Results:

• A significant proportion of patients (21.4%) had an AI ECG-age >5 years older than their chronological age.
• A larger age gap was linked to increased comorbidities (renal failure, heart failure, prior myocardial infarction) and poorer ejection fraction.
• Patients with a >5-year age gap faced higher risks of postoperative atrial fibrillation, prolonged ventilation, blood transfusion, elevated creatinine, and longer hospital stays.
• Long-term survival was significantly lower in patients with an age gap >5 years (HR 1.4, P < .001).

Conclusions:

• An elevated age gap, identified by AI ECG-age, signifies a patient cohort with greater comorbidities and advanced physiological aging.
• AI ECG-derived physiological age is an independent predictor of adverse operative outcomes and increased long-term mortality following CABG.