Available online 27 August 2022

Technological advances have led to electrocardiograph (ECG) functionality becoming increasingly accessible in wearable health devices, which has the potential to vastly expand the clinician's ability to monitor, diagnose, and manage cardiac health conditions. However, achieving the high signal quality necessary to make an accurate and confident diagnosis is inherently challenging on consumer device-acquired ECGs. Effective signal conditioning is crucial to make ECG data from wearable devices clinically actionable.

This study evaluates the heart rate (HR) performance of ECG data collected on the HeartKey® Test Watch, a single lead, dry electrode wrist wearable, against data acquired on two criterion devices: the Bittium® Faros 180, a gold standard wet electrode ambulatory monitoring device, and the HeartKey Chest Module.

ECG data was simultaneously acquired on three devices during a multi-stage protocol (sitting, walking, standing) designed to reflect the motion noise of real-life scenarios. Raw ECGs from the HeartKey Test Watch and HeartKey Chest Module were processed through HeartKey software, and the accuracy of the outputted heart rate data was compared to that of the criterion device at each stage of the protocol. A beat rejection analysis was performed to provide insight into the degree of high frequency noise present in ECGs recorded on the HeartKey Test Watch.

Data acquired on the HeartKey Test Watch and processed by HeartKey software generated HR metrics that closely matched that of the criterion devices throughout the protocol. Bland-Altman analysis showed a mean absolute HR difference of 0.74, 1.21, 0.80 bpm during the sitting, walking, and standing stages respectively, which is within the ±10% or ± 5 bpm range required by ANSI EC13. ECG data from the HeartKey Test Watch had a higher beat rejection rate relative to the HeartKey Chest Module (8.5% vs ~0%) due to the excessive high frequency noise generated during the motion-based protocol.

HeartKey software demonstrated highly accurate HR performance, comparable to that of the criterion Faros device, when processing challenging ECG data acquired on a single lead dry, electrode wrist wearable during both non-motion and motion-based protocols.

ECG

Algorithm

Signal processing

Impedance

Dry electrode

Heart rate

© 2022 Published by Elsevier Inc.