Abstract 4139236: Artificial Intelligence–Electrocardiography to Predict Incident Atrial Fibrillation and Survival Following Kidney Transplant

Background: New-onset atrial fibrillation (AF) is common among kidney transplant (KTx) recipients and is associated with reduced patient survival. Predictors of AF after KTx are not well understood, although AF can be associated with traditional and non-traditional risk factors. While artificial intelligence-enabled electrocardiography (AI-ECG) has shown promise in predicting incident AF, its predictive and prognostic implications in the KTx population have not yet been evaluated. Hypothesis: AI-ECG can predict new-onset AF and carries prognostic implications in patients undergoing KTx. Aims: To evaluate the clinical implications of applying AI-ECG to the preoperative ECGs of recipients of a KTx. Methods: Patients without a history of AF who underwent KTx at three referral centers between 2011 and 2021, with at least one preoperative ECG, were included in this retrospective study. Preoperative ECGs were analyzed using a previously developed AI-ECG algorithm to estimate the probabilities of new-onset AF. Based on these probabilities, patients were categorized into two groups: high and low probability of incident AF. The optimal cut-off value for the AI-ECG tool was determined using ROC analysis. The incidence of new-onset AF and mortality at 5 years post KTx were compared between these two groups using univariate and multivariate Cox regression analyses. Results: In total, 6246 patients were included (mean age 52.9 ±14.3 years, 58.7% males). A pre-transplant AI-ECG probability of AF >10% was identified as the most accurate cutoff point to distinguish between patients at low risk and high risk of incident AF (ROC = 0.72). The study found that a preoperative AI-ECG high risk of AF demonstrated not only a strong association with new-onset AF (HR 2.54, 95%CI 2.02-3.19, p<0.001 on multivariable analysis, Figure 1A ) but was also associated with an increased risk of mortality both on univariate (HR 2.73, 95%CI 2.34-3.18, p<0.001) and multivariate (HR 1.57, 95%CI 1.29-1.91, p<0.001 ) analysis at 5-year follow-up ( Figure 1B ). Conclusion: The application of AI-ECG to a single widely available test may help assess the risk of new-onset AF and mortality in patients undergoing KTx. Accurate risk stratification may identify a high-risk subgroup of transplant recipients who can benefit from closer AF monitoring, risk factor modification, and early management.