A Comparative Analysis of Signal Quality and Morphological Fidelity in ECG Smartwatches

Abstract

This study provides a direct and objective comparison of ECG signal quality and morphological feature preservation across four leading commercial smartwatches: the Apple Watch Series 9, Samsung Galaxy Watch 6, Fitbit Sense 2, and Withings ScanWatch, using a controlled dataset. The Philips TC30 is used as the reference standard. ECG signal quality is assessed using Signal-to-Noise Ratio (SNR), Standard Deviation of Baseline Wander, and Power Spectral Density (PSD). Preservation of morphological features is assessed by comparing the performance of our automatic J-point and ST deviation detection algorithms when applied separately to recordings from each smartwatch. J-point detection is evaluated by comparing each device's performance against the reference device, while ST deviation is assessed using simulator-defined ground truth values. The Apple Watch and Withings ScanWatch achieved the best SNR, while the Samsung Galaxy Watch showed substantially lower values. The Withings ScanWatch demonstrated the most stable baseline, whereas the Fitbit Sense 2 showed the greatest baseline variability. All devices exhibited a consistent negative bias in J-point detection, suggesting that internal signal processing tends to smooth small changes in waveforms. For ST-segment classification, the Apple Watch achieved the most balanced performance in terms of sensitivity and specificity. These results provide researchers and clinicians with data-driven guidance for selecting the most appropriate device for their specific needs.