Background:Cerebral thromboembolism during atrial fibrillation (AF) ablation is an infrequent (0.17%) complication in part owing to strict adherence to intraprocedural anticoagulation. Failure to maintain therapeutic anticoagulation can lead to an increase in events, including silent cerebral ischemia. Objective:To evaluate a computerized, clinical decision support system (CDSS) to dose intraprocedural anticoagulation and determine if it leads to improved intraprocedural anticoagulation outcomes during AF ablation. Methods:The Digital Intern dosing algorithm is an adaptive, rule-based CDSS for heparin dosing. The initial dose is calculated from the patient's weight, baseline activated clotting time (ACT), and outpatient anticoagulant. Subsequent recommendations adapt based on individual patient ACT changes. Outcomes from 50 cases prior to algorithm introduction were compared to 139 cases using the algorithm. Results:Procedures using the dosing algorithm reached goal ACT (over 300 seconds) faster (17.6 ± 11.1 minutes vs 33.3 ± 23.6 minutes pre-algorithm, P < .001). ACTs fell below goal while in the LA (odds ratio 0.20 [0.10-0.39], P < .001) and rose above 400 seconds less frequently (odds ratio 0.21 [0.07-0.59], P = .003). System Usability Scale scores were excellent (96 ± 5, n = 7, score >80.3 excellent). Preprocedure anticoagulant, weight, baseline ACT, age, sex, and renal function were potential predictors of heparin dose to achieve ACT >300 seconds and final infusion rate. Conclusion:A heparin dosing CDSS based on rules and adaptation to individual patient response improved maintenance of therapeutic ACT during AF ablation and was rated highly by nurses for usability.

Ablation in the left atrial appendage (LAA) is risky due to numerous ridges separated by thin tissue. There are few reported cases of successful ablation of incessant atrial tachycardia (AT) originating within the LAA.

Intraprocedural anticoagulation during atrial fibrillation (AF) ablation reduces the risk of cerebral thromboembolism during ablation. Heparin dosing strategies vary, can result in sub-therapeutic levels, and require frequent physician decisions if not protocol driven.

Introduction: Pulmonary veins (PV) often trigger atrial fibrillation (AF). However, the underlying mechanism is not completely understood and predicting who benefits from PV isolation (PVI) is imprecise. We propose that arrhythmogenic PV remodeling may be related to myocardial mechanical function. Flow-encoded cardiac magnetic resonance imaging (4D Flow MRI) based computational modeling enables assessment of left atrial (LA) hemodynamics and biomechanics. In this pilot study, 4D Flow MRI derived parameters were compared to invasive electroanatomic mapping (EAM) and hemodynamics to explore potential non-invasive evaluation of PV remodeling and arrhythmogenicity. Methods: Patients (5 female, 4 male, age 64±6.7 years) with paroxysmal or early persistent AF scheduled for PVI were enrolled. All patients underwent 4D Flow MRI prior to ablation (in sinus rhythm) and intraprocedural EAM (Carto) with atrial pacing under baseline conditions and after increasing LA pressure with a fluid bolus. Non-invasive measures included peak and average velocity (m/s) and net flow (ml/cycle) in each vein (n=36) and averaged for each patient. Apparent conduction velocities in the proximal, mid and distal portions of each vein were calculated under baseline and stretch conditions from EAM. Change in LA pressure per volume bolus, a measure of atrial stiffness, was compared to non-invasive parameters. Results: In patients with LA stiffness greater than median, peak velocities (0.77±0.12 vs. 0.62±0.24 m/s, p=0.26) and net flows were higher (17±2.0 vs. 14.0±2.4 ml/cycle, p=0.06). Patients with paroxysms of spontaneous arrhythmias during mapping had higher LA stiffness (11.5±5.7 vs. 7.6±5.3 mmHg/L, p=0.30). There was a significant correlation between non-invasive average velocity and apparent conduction velocity in the proximal portion of the vein (R=-0.36, p=0.03) and the suggestion of a correlation between increasing peak flow velocity and decreased apparent conduction velocity under stretch conditions compared to baseline (R=-0.24, p=0.16). Conclusions: Our findings may indicate a link between atrial hemodynamics, electrical remodeling and PV arrhythmogenicity. Additional studies are warranted to confirm and better define these findings.

Introduction: Therapeutic anticoagulation during atrial fibrillation (AF) ablation reduces the risk of cerebral thromboembolism during ablation. Heparin dosing strategies vary and can result in sub-therapeutic anticoagulation. Hypothesis: We hypothesize that a heparin dosing algorithm will improve maintenance of guideline-based therapeutic activated clotting times (ACT) during AF ablation. Methods: Starting from published guidance on heparin dosing during AF ablation, an artificial intelligence algorithm based on a decision tree and other mathematical considerations was developed. The algorithm launches directly from the electronic medical record and captures deidentified data for future improvements. The rules were modified over several months to establish the tested algorithm. The strategy utilized heparin boluses and an infusion. Dosing was adapted based on an individual patient’s response to the first heparin bolus. Goal ACT was 300-350 s. Outcomes from a month prior to algorithm introduction were compared to those from one month using the developed algorithm. The measured outcomes included time from first bolus to therapeutic ACT (>300 s), number of heparin boluses to achieve goal ACT, time in left atrium below therapeutic ACT, and ACT value when below goal in left atrium. Results: Data from seventeen pre-algorithm procedures were compared to fourteen procedures completed using the dosing algorithm. Compared to procedures using typical dosing, procedures using the dosing algorithm reached goal ACT faster (16.1 ± 5.3 min compared to 26.2 ± 14.5 min, p = 0.02) with fewer heparin boluses (1.1 ± 0.3 boluses compared to 1.6 ± 0.9 boluses, p = 0.04). The ACT did drift below goal in some cases (n = 4 using algorithm and n = 5 using typical dosing). The time below goal was not different between dosing strategies (16.5 ± 8.7 min for algorithm compared to 22.4 ± 7.9 min for typical dosing). However, when the ACT drifted below goal with the dosing algorithm, it was significantly closer to goal than with typical dosing (295 ± 3 s compared to 268 ± 21 s, p = 0.02). Conclusion: An artificial intelligence algorithm based on a decision tree and adaptation to individual patient kinetics may improve maintenance of guideline based therapeutic ACT during AF ablation.