Wrist-worn device combining PPG and ECG can be reliably used for atrial fibrillation detection in an outpatient setting

Abstract

AimsThe aim was to validate the performance of a monitoring system consisting of a wrist-worn device and a data management cloud service intended to be used by medical professionals in detecting atrial fibrillation (AF).MethodsThirty adult patients diagnosed with AF alone or AF with concomitant flutter were recruited. Continuous photoplethysmogram (PPG) and intermittent 30 s Lead I electrocardiogram (ECG) recordings were collected over 48 h. The ECG was measured four times a day at prescheduled times, when notified due to irregular rhythm detected by PPG, and when self-initiated based on symptoms. Three-channel Holter ECG was used as the reference.ResultsThe subjects recorded a total of 1,415 h of continuous PPG data and 3.8 h of intermittent ECG data over the study period. The PPG data were analyzed by the system’s algorithm in 5-min segments. The segments containing adequate amounts, at least ~30 s, of adequate quality PPG data for rhythm assessment algorithm, were included. After rejecting 46% of the 5-min segments, the remaining data were compared with annotated Holter ECG yielding AF detection sensitivity and specificity of 95.6 and 99.2%, respectively. The ECG analysis algorithm labeled 10% of the 30-s ECG records as inadequate quality and these were excluded from the analysis. The ECG AF detection sensitivity and specificity were 97.7 and 89.8%, respectively. The usability of the system was found to be good by both the study subjects and the participating cardiologists.ConclusionThe system comprising of a wrist device and a data management service was validated to be suitable for use in patient monitoring and in the detection of AF in an ambulatory setting.Clinical Trial Registration: ClinicalTrials.gov/, NCT05008601.

Ambulatory Atrial Fibrillation Detection and Quantification by a wrist-worn AI-device Compared to Standard Holter Monitoring

“…Reporting of PPV was less frequent in the existing literature. In an ambulatory setting, the PPV ranged from 39.9% to 97.7% when considering both PPG-based and single-lead ECG-based algorithms [24,26,27]. Both CW2's algorithms, with 81% and 92%, respectively, are again on the high end of this spectrum.…”

“…Furthermore, as the CW2 can measure multiple other parameters (including pulse rate, SpO2, breathing frequency, non-invasive blood pressure, and accelerometer data), early detection and monitoring of various (cardiological) health conditions is enabled. These qualities set the CW2 apart from similar AF-detection devices using a PPG-base or single-lead ECG-based algorithm, which focuses solely on AF detection and/or consists of multiple or large wearable devices [26,27]. Moreover, the CW2's high accuracy and ease of use make it an ideal alternative for AF screening and monitoring, offering a less burdensome option compared to standardof-care methods whilst allowing an extended monitoring period.…”

ECG signal quality in intermittent long-term dry electrode recordings with controlled motion artifacts

“…Stainless steel appears to be a low-noise material by itself 78 and on the skin, as found in the present and previous studies 14 , 23 . A wrist device featuring stainless steel dry electrodes for recording intermittent long-term ECG has recently been validated for atrial fibrillation detection 79 . In this application, the test subjects were stationary during the recordings, which averts the issue of a lowered SNR during movement.…”

Artificial intelligence for atrial fibrillation detection, prediction, and treatment: A systematic review of the last decade (2013–2023)

“…Liao et al (2022) tested a SVM and a random forest classifier to detect AF from a wrist PPG. Saarinen et al (2023) validated a wrist-worn device for AF detection using PPG and intermittent ECG compared with Holter ECG. Lahdenoja et al (2018) and Navoret et al (2013) both applied signal processing and ML techniques to accelerometer and gyroscope signals recorded from smartphones to extract features related to heartbeat timing, entropy measures, and signal complexity, to train classifiers like SVM to detect AF.…”