Atrial Fibrillation Risk Score by Artificial Intelligence ECG Predicts 12-Month Recurrence of Atrial Fibrillation After Catheter Ablation

The atrial fibrillation (AF) risk score generated by artificial intelligence electrocardiography (SmartECG-AF) has previously been validated and shown to have favorable diagnostic performance. This study evaluated longitudinal changes in the SmartECG-AF risk score after AF catheter ablation (CA) and assessed whether the score predicts AF recurrence within 12 months after ablation. This study included 144 patients who underwent AF CA in 2022, including 112 males with a mean age of 61.6 ± 10.76 years. The 1-year follow-up rate was 90.3%. AF recurrence was assessed over 12 months using multiple electrocardiogram-based monitoring modalities, and clinical outcomes were analyzed according to the SmartECG-AF risk score. Hypertension, diabetes mellitus, cerebrovascular accident, heart failure, and vascular disease were present in 37.5%, 18.1%, 6.9%, 13.2%, and 2.8% of patients, respectively. The mean CHA2DS2-VASc score was 1.5 ± 1.20, and the mean creatinine level was 0.93 ± 0.170 mg/dL. Persistent AF was present in 97 patients (67.4%), and 47 patients had paroxysmal AF. Radiofrequency CA and cryoballoon CA were performed in 81.3% and 18.8% of cases, respectively. The 12-month AF-free survival rate was 65.4%. Patients with AF recurrence had a larger left atrial diameter (47.2 ± 6.40 vs. 43.4 ± 6.13 mm, P = 0.002) and more frequent early recurrence during the 3-month blanking period (62.5% vs. 17.3%, P < 0.001). The SmartECG-AF risk score decreased overall after CA (84.2 ± 24.15 to 76.1 ± 29.45, P = 0.007); however, changes from pre-CA to post-CA and 3 months were not significantly associated with recurrence. The pre-CA risk score was higher in patients with AF recurrence (92.5 ± 14.04 vs. 81.7 ± 25.61, P = 0.005). A higher pre-CA risk score independently predicted 12-month AF recurrence after adjustment for sex, age, body mass index, and left atrial diameter (hazard ratio, 1.045; confidence interval, 1.005–1.086). The SmartECG-AF risk score decreases after CA, but the magnitude of reduction is not associated with AF recurrence. A higher pre-CA risk score predicts poorer outcomes after AF CA.