P4257Assessing the accuracy of an automated atrial fibrillation detection algorithm using novel smartphone technology

Abstract

Background: The Alivecor cardiac monitor (ACM) system is a two-electrode handheld cardiac rhythm smartphone recorder paired with a rhythm adjudication algorithm. The automated algorithm interprets a 30 second rhythm strip as "Normal", "Possible AF detected", or "Unreadable." Purpose: The primary objective was to examine whether the ACM algorithm can accurately differentiate sinus rhythm from AF compared to physician interpreted 12-lead ECGs and ACM recordings. The secondary objective was to assess patient feedback regarding ACM. Method(s): Consecutive patients admitted for antiarrhythmic drug initiation for AF were provided with ACMs. 12-lead ECGs were taken twice daily. Patients performed ACM recordings immediately following each ECG. Cardiac rhythm on 12- lead ECGs and ACM recordings were interpreted by blinded electrophysiologists. Sensitivity, specificity and the kappa coefficient were assessed. Patients were provided a survey assessing feedback regarding the ACM. Result(s): Fifty-two patients were enrolled (mean age 68, 67% male). There were 225 paired ACM recordings and ECGs. The algorithm interpretation was missing or labeled as non-interpretable in 62 (27.5%) of recordings for multiple reasons (truncated recording, noise, slow heart rate, other). When the algorithm did not provide a diagnosis, blinded electrophysiologists were able to provide interpretation in 92% of these recordings. After exclusion of non-interpretable ACM recordings, automated ACM had a sensitivity of 96.6% and specificity of 94% for the detection of AF when compared to physician interpreted ECGs, with a kappa coefficient of 0.89. The majority of patients (93.6%) found the ACM easy to use, and 59.6% noted that use lessened AF-diagnosis related anxiety. 63.8% of survey respondents preferred continued use of the ACM for AF detection. Conclusion(s): Automated ACM adjudication, when possible, demonstrated very good accuracy when compared to physician interpreted simultaneous ECGs. Patients found the ACM system easy to use, and many felt its utilization could improve stress pertaining to AF.