P151. Prognostic Value of AI ECG–Derived Diastolic Function in Surgical Aortic Valve Replacement
Presented During: Adult Cardiac Poster Session II
Mayo Clinic - Rochester
Rochester, MN
United States - Contact Me
I am a postdoctoral research fellow at Mayo Clinic, working under the mentorship of Dr. Hartzell Schaff. My research primarily focuses on the outcomes of transaortic and transapical septal myectomy for obstructive hypertrophic cardiomyopathy, the outcomes of mitral and tricuspid valve surgeries (with the focus on carcinoid heart disease), and the integration of artificial intelligence in surgical practice.
I am deeply committed to pursuing my dream of becoming an academic surgeon, with my immediate goal being to enter a surgery residency before specializing in cardiothoracic surgery. Research is a core aspect of my professional identity, and I am determined to incorporate it into my future clinical practice, running studies that can enhance and advance the field of surgery for the benefit of both patients and fellow surgeons.
Monday, May 4, 2026: 9:00 AM - 4:00 PM
McCormick Place Lakeside Center
Room: Exhibit Hall, Poster Area
Description
Background: Diastolic dysfunction is common in patients with aortic stenosis (AS) and may influence short and long-term outcomes following surgical aortic valve replacement (sAVR). While assessment of diastolic function by echocardiography is challenging, AI ECG diastolic function assessment can be accurate and readily accessible. We hypothesized that AI-generated diastolic function grades based on perioperative ECGs strongly correlate with early and late outcomes following sAVR replacement.
Methods: We analyzed 5,503 patients undergoing isolated sAVR between 1993 and 2023. Diastolic function was assessed using a validated deep-learning AI model applied to 12-lead ECGs done preoperatively and on postoperative follow-up. Diastolic function grades were classified by AI into Grades 1 to 3. Longitudinal trend analyses and multivariable regression models were used to assess study endpoints.
Results: Among 5,503 patients (mean age 72.4 ± 10.8 yr; 39% female), more abnormal diastolic function, as indicated by higher AI-ECG diastolic function grades, was associated with greater comorbidity burden, including diabetes, renal disease, and heart failure. Compared to patients with AI-ECG diastolic function Grade 1, those with Grade 2 and 3 were at an independently higher risk of in-hospital mortality (OR 1.7, p=0.06; OR 2.5, p=0.007, respectively), prolonged mechanical ventilation (OR 1.4, p=0.016; OR 2.0, p<0.001), renal failure (OR 1.7, p=0.031; OR 2.4, p=0.002), and postoperative transfusion (OR 1.4, p<0.001; OR 1.5, p<0.001).
At 5-year follow-up, patients with Grade 3 diastolic function showed the least improvement in postoperative diastolic function (all p<0.001) (Figure 1A). Additionally, Grades 3 diastolic function at baseline was independently associated with the highest late mortality (HR 2.45; p < 0.001) (Figure 1B). Notably, lack of improvement in AI-ECG diastolic grade within the first year following surgery was independently associated with increased late mortality (HR 1.4, p<0.001) (Figure 1C)
Conclusion: Higher preoperative AI-ECG diastolic grades were associated with increased early postoperative complications, reduced long-term survival, and less favorable improvement in diastolic function over time. Lack of 1-year improvement in AI-ECG grade is also independently associated with increased late mortality following sAVR. Thus, assessment of diastolic function by AI-ECG offers a robust tool for risk stratification and monitoring of AS patients
Authors
Tedy Sawma (1), Hartzell Schaff, MD (2), Sina Danesh (2), Arman Arghami (3), Eunjung Lee (2), Masoomeh Aslahishahri (2), John Stulak (2), Kevin Greason (2), Joseph Dearani (4), Francisco Lopez-Jimenez (2), Paul Friedman (2), Sorin Pislaru (2), Itzhak Zachi [email protected] (2), Jae Oh (2), Juan Crestanello (5)
Institutions
(1) Mayo Clinic - Rochester, Rochester, MN, (2) Mayo Clinic, Rochester, MN, (3) Mayo Clinic Hospitals Rochester, MN, Rochester, MN, (4) Mayo Clinic College of Medicine and Science, Rochester, MN, (5) Mayo Foundation, Rochester, MN
Methods: We analyzed 5,503 patients undergoing isolated sAVR between 1993 and 2023. Diastolic function was assessed using a validated deep-learning AI model applied to 12-lead ECGs done preoperatively and on postoperative follow-up. Diastolic function grades were classified by AI into Grades 1 to 3. Longitudinal trend analyses and multivariable regression models were used to assess study endpoints.
Results: Among 5,503 patients (mean age 72.4 ± 10.8 yr; 39% female), more abnormal diastolic function, as indicated by higher AI-ECG diastolic function grades, was associated with greater comorbidity burden, including diabetes, renal disease, and heart failure. Compared to patients with AI-ECG diastolic function Grade 1, those with Grade 2 and 3 were at an independently higher risk of in-hospital mortality (OR 1.7, p=0.06; OR 2.5, p=0.007, respectively), prolonged mechanical ventilation (OR 1.4, p=0.016; OR 2.0, p<0.001), renal failure (OR 1.7, p=0.031; OR 2.4, p=0.002), and postoperative transfusion (OR 1.4, p<0.001; OR 1.5, p<0.001).
At 5-year follow-up, patients with Grade 3 diastolic function showed the least improvement in postoperative diastolic function (all p<0.001) (Figure 1A). Additionally, Grades 3 diastolic function at baseline was independently associated with the highest late mortality (HR 2.45; p < 0.001) (Figure 1B). Notably, lack of improvement in AI-ECG diastolic grade within the first year following surgery was independently associated with increased late mortality (HR 1.4, p<0.001) (Figure 1C)
Conclusion: Higher preoperative AI-ECG diastolic grades were associated with increased early postoperative complications, reduced long-term survival, and less favorable improvement in diastolic function over time. Lack of 1-year improvement in AI-ECG grade is also independently associated with increased late mortality following sAVR. Thus, assessment of diastolic function by AI-ECG offers a robust tool for risk stratification and monitoring of AS patients
Authors
Tedy Sawma (1), Hartzell Schaff, MD (2), Sina Danesh (2), Arman Arghami (3), Eunjung Lee (2), Masoomeh Aslahishahri (2), John Stulak (2), Kevin Greason (2), Joseph Dearani (4), Francisco Lopez-Jimenez (2), Paul Friedman (2), Sorin Pislaru (2), Itzhak Zachi [email protected] (2), Jae Oh (2), Juan Crestanello (5)
Institutions
(1) Mayo Clinic - Rochester, Rochester, MN, (2) Mayo Clinic, Rochester, MN, (3) Mayo Clinic Hospitals Rochester, MN, Rochester, MN, (4) Mayo Clinic College of Medicine and Science, Rochester, MN, (5) Mayo Foundation, Rochester, MN
Presentation Duration
There is no formal presentation for posters. Your poster will be on display on your assigned day from 9:00AM - 4:00PM
View Submission
ADULT CARDIAC
Aortic Valve