Technical characterization of the single-lead electrocardiogram signal from four different smartwatches and its clinical implications

Abstract Funding Acknowledgements Type of funding sources: Public Institution(s). Main source(s): Department Cardiology, University Hospital Basel, Switzerland Institute for Medical Engineering and Informatics, School Life Sciences, Applied Sciences Arts Northwestern Switzerland, Muttenz, Background Smartwatches with the capability single-lead electrocardiograms (ECGs) acquisition have been established as valuable tool detection atrial fibrillation (AF). For currently available smartwatches, however, technical specification acquired ECG signal is unclear. Purpose The aim current study was to characterize quantitatively raw four smartwatches (Apple watch series 6, Withings ScanWatch, Samsung Galaxy Watch 4, Fitbit Charge 5) assess their clinical applicability based on features. Method A bench test performed applying a synthetic, well-characterized electrodes watches (Figure: (A) Apple (B) (C) (D) 5). frequency response this input Bode plots were calculated (Figure, second column: diagrams show (upper) phase angles (lower)). To compare characteristics watches, synthetical pre-recorded surface dataset determined above third shown in grey, blue). Results responses all shows bandpass filter, an additional 50/60 Hz narrow-band filter (Figure). While recommended high-pass cut-off at 0.5 fulfilled by devices (1), low-pass observed below 70Hz (instead 150Hz) resulting significantly damped signal. In detail, damping -6dB, corresponding amplitude reduction 50%, 70 0.4 watches. QRS complex Apple, watch. pronounced noise filtering visible Smartwatch. difference clinically relevant characteristic fiducial points, such onset P-wave or complex, cannot be these examples. However, when using tangent method detect end T-wave calculate QTc value, between measures can Watch. Conclusion terms from commercially differ must taken into consideration attempting use beyond AF.