AI Model Identifies Age Cutoff for Personalized Blanking Period Post Atrial Fibrillation Ablation

Ala Assaf, N/A – Research Fellow, Tulane University School of Medicine; Ghassan Bidaoui, N/A – Research Fellow, Tulane University School of Medicine; Mayana Bsoul, N/A – Research Fellow, Tulane University School of Medicine; Nadia Chamoun, N/A – Research Fellow, Tulane University School of Medicine; Nour Chouman, N/A – Research Fellow, Tulane University School of Medicine; Alexander Dahl, N/S – Medical Student, Tulane University School of Medicine; Han Feng, PhD – Biostatistician, Tulane University School of Medicine; Yishi Jia, N/A – Biostatistician, Tulane University School of Medicine; Yingshuo Liu, N/A – Biostatistician, Tulane University School of Medicine; Emma Lunn, N/A – Medical Student, Tulane University School of Medicine; Nassir Marrouche, M.D. – Professor of Medicine, Tulane University School of Medicine; Cecile McGarvey, n/a – Medical Student, Tulane University School of Medicine; Amitabh Pandey, M.D. – Assistant Professor, Tulane University School of Medicine; Swati Rao, M.D. – Assistant Professor, Tulane University School of Medicine; Susan Tang, N/A – Biostatistician, University of Southern California; Francisco Tirado, M.D. Ph.D – Faculty, Tulane University School of Medicine; Hadi Younes, N/A – Research Fellow, Tulane University School of Medicine

Purpose: The clinical efficacy of using the arbitrary 90-day blanking period following atrial fibrillation (AF) ablation has recently been questioned. A 33-day optimal blanking period has been previously proposed, but it has not been investigated if this blanking period is equally appropriate for all patient groups. We sought to personalize blanking period after ablation. We will leverage an artificial intelligence (AI) model that determines characteristics of patients who may benefit from longer or shorter blanking period times.

Material and Methods: We used causal tree learning to identify heterogeneous subpopulations that respond differently to the prespecified 33-day blanking period. We used the grid search method for each identified subpopulation, examining 27 variables including demographics, medication history, comorbidities and MRI data, to discover its optimal blanking period. The outcomes were compared through Kaplan-Meier (KM) curves using time to AF recurrence following 90 days as the outcome. Baseline statistics were compared through the Wilcoxon and Chi-square tests for continuous and categorical variables, respectively.

Results: In total 688 patients from the multicenter prospective clinical trial DECAAF II with blanking period information collected were considered in the study. Based on the causal tree model results, the whole population is divided into 2 groups. Group 1 contained 295 patients with their age < 61 years and group 2 contained 393 patients with their age ≥61 years. The optimal blanking periods were found to be 24 and 40 days for group 1 and group 2.

Conclusions: This heterogenous blanking period, during which post ablation AF recurrences are not indicative of long-term procedural success, may aid in more precise decision-making regarding re-ablation. Using an AI model, we are able to assign 2 different blanking periods with individuals over 61 years having a longer blanking period by 16 days, possibly due to slower healing process and scar formation post ablation.