Cited by 245ECG Monitoring and AnalysisNon-Invasive Vital Sign MonitoringAtrial Fibrillation Management and Outcomes

Smartwatch Performance for the Detection and Quantification of Atrial Fibrillation

Wasserlauf, Jeremiah, You, Cindy, Patel, Ruchi, Valys, Alexander, Albert, David, Passman, Rod·2019·Circulation: Arrhythmia and Electrophysiology

Abstract

Background: Atrial fibrillation (AF) burden and duration appear to be related to stroke risk. A wearable consumer electronic device could provide long-term assessment of these measures inexpensively and noninvasively. This study compares the accuracy of an AF-sensing watch (AFSW; Apple Watch with KardiaBand) with simultaneous recordings from an insertable cardiac monitor (ICM; Reveal LINQ). Methods: SmartRhythm 2.0, a convolutional neural network, was trained on anonymized data of heart rate, activity level, and ECGs from 7500 AliveCor users. The network was validated on data collected in 24 patients with ICMs and a history of paroxysmal AF who simultaneously wore the AFSW with SmartRhythm 0.1 software. The primary outcome was sensitivity of the AFSW for AF episodes ≥1 hour. Secondary end points included sensitivity of the AFSW for detection of AF by subject and sensitivity for total AF duration across all subjects. Subjects with >50% false-positive AF episodes on ICM were excluded. Results: We analyzed 31 348.9 hours (mean (SD), 11.3 (4.4) hours/day) of simultaneous AFSW and ICM recordings in 24 patients. The ICM detected 82 episodes of AF ≥1 hour while the AFSW was worn, with a total duration of 1127.1 hours. Of these, the SmartRhythm 2.0 neural network detected 80 episodes (episode sensitivity, 97.5%) with a total duration of 1101.1 hours (duration sensitivity, 97.7%). Three of the 18 subjects with AF ≥1 hour had AF only when the watch was not being worn (patient sensitivity, 83.3%; or 100% during time worn). Positive predictive value for AF episodes was 39.9%. Conclusions: An AFSW is highly sensitive for detection of AF and assessment of AF duration in an ambulatory population when compared with an ICM. Such devices may represent an inexpensive, noninvasive approach to long-term AF surveillance and management.

Integrity signals

Registry

Crossref · 10.1161/circep.118.006834

Cites

28 of 28 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Claims the paper makes

This paper’s full text isn’t openly available, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Authors
Wasserlauf, J., You, C., Patel, R., Valys, A., Albert, D., & Passman, R.
Affiliations
  • Northwestern University (US)
  • AliveCor, Mountain View, CA (A.V., D.A.).
Venue
Circulation: Arrhythmia and Electrophysiology
Published
2019-06
Volume
12
Issue
6
Pages
e006834-e006834
ISSN
  • 1941-3084 (ISSN-L)
  • 1941-3149
OpenAlex
W2946078250
Cited by
245
References
28
Citations by year
2026: 10 · 2025: 29 · 2024: 30 · 2023: 40 · 2022: 32 · 2021: 59 · 2020: 35 · 2019: 9 · 2018: 1
FWCI
24.38
Type
journal article
Language
en
Open access
closed
Locations
  • Circulation Arrhythmia and Electrophysiology (publishedVersion)
  • PubMed (publishedVersion)
Topics
  • Health Sciences › Medicine › Cardiology and Cardiovascular Medicine › ECG Monitoring and Analysis
  • Physical Sciences › Engineering › Biomedical Engineering › Non-Invasive Vital Sign Monitoring
  • Health Sciences › Medicine › Cardiology and Cardiovascular Medicine › Atrial Fibrillation Management and Outcomes
Keywords
atrial fibrillation, smartwatches, wearable ECG, Apple Watch, implantable cardiac monitors, convolutional neural networks, atrial fibrillation detection, atrial fibrillation duration, paroxysmal atrial fibrillation, long-term cardiac monitoring, stroke risk
MeSH terms
Wearable Electronic Devices, Deep Learning, Action Potentials, Aged, Atrial Fibrillation, Female, Heart Atria, Heart Rate, Humans, Male, Predictive Value of Tests, Signal Processing, Computer-Assisted, Reproducibility of Results, Electrocardiography, Ambulatory, Remote Sensing Technology, Mobile Applications

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review