Level Accuracy of Automatic and Real Time Detection of Atrial Fibrillation with A New Wireless ECG Recorder (The SmartCardia)

Background: Atrial fibrillation (AF) is the most common cardiac arrhythmia but is currently under-diagnosed since it can be asymptomatic.Early detection of AF could be highly beneficial for the prevention of stroke, which is a major risk associated with AF, with a fivefold increase.The advent of portable monitoring devices can help uncover the underlying dynamics of human health in a way that has not been possible before. Method:The purpose of this study was to validate the automated analysis of AF by the Smart-Cardia's proprietary health monitoring device (ScaAI patch, SmartCardia S.A., Lausanne, Switzerland).To this end, a model was created and tested on three publicly available databases comprised of 243,960 ECG segments of 30-seconds.The model was further tested against a set of 500 ECG streams of 30-seconds (recorded by ScaAI patch -across different clinical trials; annotated by 3 different cardiologists), especially representing problematic conditions, when determination the underlying rhythm was challenging. Results:The created model obtained F1-scores of 94.42 against a test set from two published available databases, and an F1-score of 92.61 (average F1-scores w.r.t each cardiologist) on the SmartCardia assembled database. Conclusion:We demonstrated that the new wireless ScalAl patch had a high capacity to automatically detect AF when compared with public database.Further studies will help identify the optimal role of the the ScaAl patch in the management of cardiac arrhythmias.