Abstract Body (Do not enter title and authors here): Background
Atrial fibrillation (AF) is the most common arrhythmia, impacting 38 million people worldwide. Oral antiarrhythmic drugs (AADs) reduce symptoms and improve cardiovascular outcomes. Inpatient monitoring is required for the AADs dofetilide and sotalol. There is no available software system designed to support safe AAD titration irrespective of patient setting. Limited availability of inpatient resources restricts access to mortality-reducing AADs, particularly for underserved populations.

Hypothesis
Machine learning (ML) algorithms can accurately measure corrected QT (QTc) interval from mobile ECG data and recommend appropriate AAD dosing comparable to electrophysiologist (EP).

Aims
To assess accuracy of ML algorithms for recommending patient-specific AAD dosing, compared to EP physician dosing decisions.

Methods
A de-identified dataset of mobile 6L mECG (mECG; KardiaMobile 6L, AliveCor) was collected from 702 patients receiving care in a congenital Long QT Clinic. A proprietary ML algorithm (SafeBeat Rx, San Francisco, CA) analyzed heart rhythm intervals and calculated QTc (“ML QTc”) for each mECG. The mECG dataset was also reviewed by EP cardiologists to manually measure QTc values (“gold standard QTc”). Basic ECG parameters (HR, RR interval, QTc) were calculated using values generated from ML and physician-annotated ECG waveforms. Based on the FDA drug label, starting doses of dofetilide and sotalol were simulated, assuming normal renal function, by both the EP cardiologist and the ML algorithm.

Results
ECG statistics are summarized in Table 1. Overall, 13.5% of patients had prolonged QTc (≥460 ms). The absolute mean differences between HR, RR interval, and ML vs gold standard QTc values were 0.5 ± 2.3 BPM, 32.9 ± 123.7 ms, and 39.4 ± 44.5 ms, respectively. The ML- and EP-chosen simulated baseline AAD dosing were equivalent for 76.0% of the patients for both dofetilide and sotalol.

Conclusion
The software-recommended AAD drug dosing utilizing the ML QTc measurement was in high agreement with EP cardiologist-determined AAD drug dosing. These data demonstrate this software platform's potential to match clinician decisions for AAD dosing, offering the ability to support streamlined AAD titration regardless of patient setting.