Po-03-218 cloud based artificial intelligence powered lesion analysis to assess progression of trainee catheter handling ability

Detailed and automated data stored by an ablation mapping system may be a valuable adjunctive training tool. To assess whether analysis of lesion level ablation parameters, through a cloud-based data curation and storage system (CARTONET), can capture progression of manual catheter handling skills in a cohort of fellows progressing through a Clinical Cardiac Electrophysiology (CCEP) training program. AF ablations involving year 1 (Y1) or year 2 (Y2) CCEP fellows between August 2020 and June 2021 were included. Lesions were tagged with ablation detail during the cases (VISITAG), and assigned to an anatomic region by an artificial intelligence powered algorithm based on machine learning. Lesions assigned to the left (L) or right (R) wide-area circumferential ablation (WACA) were analyzed. Data from the first (Q1) and last (Q4) quartile of procedures, and across all procedures, for each fellow in Y1 and Y2 were compared to assess progression of operator proficiency. Y1 fellows demonstrated improved catheter stability between Q1 and Q4 during L WACA (Q4 1.10 ± 0.74 vs 1.29 ± 0.81 Q1, p=<0.0001) and R WACA (Q4 0.90 ± 0.64 vs 1.17 ± 0.81 Q1, p<0.0001)) lesions while Y2 fellows demonstrated no significant change (L: Q4 1.19 ± 0.82 vs 1.15 ± 0.77 Q1, p = 0.2561; R: Q4 0.99 ± 0.70 vs 0.99 ± 0.69 Q1, p = 0.9644). The overall proportion of low contact force (CF) L WACA and R WACA lesions delivered by Y1 fellows was greater than the proportion delivered by Y2 fellows during the year (L: Y1 1.25 ± 2.42 vs 0.60 ± 1.30% Y2, p=0.0343; R: Y1 1.28 ± 1.95% vs 0.47 ± 1.07% Y2, p=0.0010). Catheter handling proficiency for lesion delivery during AF ablation in Y1 fellows improved between Q1 and Q4, resulting in improved catheter stability and fewer low CF lesions. During Y2 these parameters remained stable suggesting catheter handling proficiency had reached a plateau. Objective data reflecting catheter handling may be valuable to assist planning of training programs in order to deliver focused training and may in the future be used to assess competency.