Artificial Intelligence 12 Lead ECG Based Heart Age Estimation and 1-year Outcomes After Heart Transplantation

Abstract

<h3>Purpose</h3> Donor-recipient age differences are known to play a role in post-transplant outcomes with donors >50 or recipients >60 resulting in more complications. Studies to date, however, have not utilized artificial intelligence (AI) based analysis to estimate heart-age in recipients after transplantation. We evaluated the impact of donor heart-age assessment by 12 lead ECG on predicting outcomes within the first year after transplantation with a validated AI model. <h3>Methods</h3> All heart transplant recipients from May 2018 to June 2020 at Mayo Clinic Florida were included. AI based ECG age analysis was performed at time of discharge and 12 months after transplantation. Change in AI ECG heart age was compared to donor and recipient age over time and analyzed for association with rejection or death within the first year after transplant. <h3>Results</h3> A total of 67 patients underwent heart transplantation during our review period. Two patients died prior to hospital discharge and were not included in the analysis. Of the remaining 65 recipients, 45 were male (69%). Median recipient age at transplant was 58 years (IQR 47-64) with a median donor age at transplant of 29 years (IQR 22 -37), p<0.01. Median donor age was 29 years younger than recipient at transplant. One-year survival was 95% with 11 patients treated for rejection during the first year after transplant. The median AI ECG predicted heart age at discharge was 50 years (IQR 41 - 60), which was associated with 1-year survival (p<0.001) and an AUC of .789. At 1 year follow up the AI ECG heart age found 33 patients age had increased to match the recipient age at transplant compared to 27 patients whose AI ECG heart age remained between the donor and recipient age (p = 0.012). Comparing younger (<60) to older recipients (>=60) at transplant correlated with younger AI ECG heart age between the two groups (46 vs 66, p < 0.01) on discharge and at 1 year (50 vs. 61, p<0.01). No significant difference in rejection (p=0.44) or death (p=0.31) between younger or older recipients at 1 year. <h3>Conclusion</h3> 12 lead ECG is a simple noninvasive procedure that can be done easily after transplantation. Our data suggests there may be a signal with AI ECG heart age and one-year endpoints. In an age of publicly reported outcomes, identification of higher risk patients may change practice patterns and improve survival. Larger studies are warranted to evaluate this finding.