Using an Artificial Intelligence ECG-Derived Biological Age for Risk Stratification of Patients Undergoing Surgical Aortic Valve Replacement

Artificial intelligence using a convolutional neural network can estimate patients' age from ECG data. We aimed to assess the prognostic value of AI ECG-derived age in predicting outcomes following surgical aortic valve replacement (SAVR). Preoperative ECGs obtained within 30 days before surgery were used to calculate AI ECG-derived age in 5,503 patients undergoing SAVR. Age gap was defined as AI ECG-derived age minus chronological age. Associations between age gap and postoperative outcomes were evaluated. A threshold of age gap >5 years was used to define advanced ECG age. Median chronological age was 73.2(65.3-79.5) years, and AI ECG-derived age was 70.4(64.6-74.8) years. Median age gap was -3.0(-8.4 to 2.5) years. In 35.7% of patients, AI ECG-derived age exceeded chronological age, and in 16.7% it was >5 years. Patients with advanced ECG age were younger but had higher rates of diabetes, renal failure, and current smoking and had higher BMI. Postoperatively, advanced ECG age was associated with increased risk of 30-day mortality (OR 1.69, p=0.043), blood transfusion (OR 1.44, p<0.001), gastrointestinal complications (OR 1.83, p=0.022), prolonged ventilation (OR 1.41, p=0.047), higher postoperative creatinine levels (B +0.09 mg/dL, p=0.014), longer ICU stay (B +13.3 hours, p=0.001), and longer hospital stay (B +0.7 days, p=0.007). Advanced ECG age was also independently associated with increased long-term mortality (HR 1.25, p=0.003). AI ECG-derived age identifies a subgroup of SAVR patients with advanced physiological aging. An elevated age gap is independently associated with worse early postoperative outcomes and reduced long-term survival following SAVR.