Artificial Intelligence Identification of Heart Failure With Preserved Ejection Fraction Substrate in Cardiac Surgery Patients

<h2>Abstract</h2><h3>Background</h3> Outcomes for patients with heart failure with preserved ejection fraction (HFpEF) who are undergoing cardiac surgery are not clear. We sought to examine the impact of artificial intelligence (AI) –diagnosed HFpEF on cardiac surgical outcomes in low- to moderate–surgical risk patients. <h3>Methods</h3> Adult patients undergoing isolated aortic valve replacement (AVR), mitral valve repair (MVr), or coronary artery bypass grafting (CABG) between 2004 and 2022 were retrospectively scored for the likelihood of HFpEF by using a commercially available echocardiogram-based AI algorithm. Patients with a left ventricular ejection fraction (LVEF)<50% were classified as having a reduced ejection fraction (HFrEF). On the basis of AI probabilities ranging from 0 to 1, patients with an LVEF ≥50% were stratified to normal function (0-0.49), moderate-probability HFpEF (0.5-0.74), and high-probability HFpEF (≥0.75). <h3>Results</h3> Among 1882 patients, 86.6% (n = 1629) had an LVEF ≥50%; of those patients, 36.8% (n = 599) were in the high-probability HFpEF group (median LVEF, 60% [interquartile range {IQR}, 56%-65%]), 6.7% (n = 109) were in the moderate-probability HFpEF group (LVEF, 61% [IQR, 59%-65%]), and 56.5% (n = 921) had normal function (LVEF, 63% [IQR, 59%-66%]). The remaining 13.4% (n = 253) were in the HFrEF group (LVEF, 40% [IQR, 31%-45%]). Compared with normal function, high-probability HFpEF was associated with increased operative mortality (1.3% vs 0.3%; <i>P</i> = .002) and 30-day readmission (12.8% vs 6.7%; <i>P</i> < .001). Over a median of 5.8 years (range, 0-20.0 years) of follow-up, high-probability HFpEF had increased mortality, HF admission, and atrial fibrillation (<i>P</i> < .01 for all) compared with normal function. Risk-adjusted all-cause mortality was greater in high-probability HFpEF (hazard ratio, 1.84; 95% CI: 1.35-2.52) vs normal function. <h3>Conclusions</h3> Using AI, we can identify patients with high-probability HFpEF at increased risk of adverse events. Identifying this subgroup may enable surgical teams to improve short- and long-term outcomes through guideline-directed medical therapy and warrants further study.