TAP reading note: Validation of nocturnal resting heart rate and heart rate variability in consumer wearables
Five wearables showed differing agreement with a chest-strap ECG reference for nocturnal HRV. Reported concordance correlation coefficients were .87 for Garmin, .97 for Oura 3, .99 for Oura 4, .82 for Polar and .94 for WHOOP. This assesses basic nocturnal measures, not manufacturers' composite recovery scores.
Updated 2026-10-29 · Archived
This is the text and publication record captured for this edition. Later corrections appear in subsequent editions. Website design and interactive software are not archived here.
What the study found
Five wearables showed differing agreement with a chest-strap ECG reference for nocturnal HRV. Reported concordance correlation coefficients were .87 for Garmin, .97 for Oura 3, .99 for Oura 4, .82 for Polar and .94 for WHOOP. This assesses basic nocturnal measures, not manufacturers' composite recovery scores.
Scope & limitations
13 healthy adults contributing 536 nights; valid nights differed by device and are not independent participant counts.
Sampling windows and algorithms differ, and nights are repeated within people. Table 3 reports Polar bias of −4.65 ms but limits of −14.30 to 23.60 ms, whose midpoint is +4.65 ms. The original PDF shows the same values; difference direction remains unresolved, so these limits should not define a precise device threshold.
Population & sample
13 healthy adults contributing 536 nights; valid nights differed by device and are not independent participant counts.
Exposure or intervention
Garmin, Oura 3/4, Polar and WHOOP compared with Polar H10 single-lead ECG processed in Kubios, allowing less than 5% corrected beats in the reference.
Outcomes
Agreement, mean bias and limits of agreement for nocturnal resting heart rate and HRV.
Time frame
Repeated nights, with device-specific valid counts of 150, 470, 139, 206 and 289.
Research funding
US Air Force Research Laboratory, award S-194-11T-001.
Research conflicts of interest
The authors declare no competing interests.
Reading scope
Full-text participants, processing, statistics, Table 3 and limitations, including visual comparison with the original PDF. Within-person agreement and difference direction require further checking.
References
Publication record
TAP-RDN-0055 · v1.0 · English
Author: TAP Editorial Team ·
The Allostasis Project is founded and supported by Resilio.
Record for v1.0 · Updated
Editorial responsibility: TAP Editorial Team
Scientific review
Not yet scientifically reviewed
Editorial note
Under the editorial responsibility of the TAP Editorial Team, this publication used AI assistance for content organization and editing. Scientific review is recorded separately on this page.
Download detailed editorial record (JSON) ↓
How responsibilities and review are recorded ↗© The Allostasis Project. All rights reserved. Reproduction or adaptation requires separate permission, subject to applicable legal exceptions.
Download complete edition record (JSON) ↧
Record integrity & available editions
SHA-256: 6ac8b0aa697775c285c10460879df500ecba4c5a9262457467b33050d7888ed4
Captured: 2026-10-29T10:31:13.635Z
- v1.0 · 2026-10-29