Paperdoi:10.1161/circ.154.suppl_1.11107

Sleep Architecture Independently Predicts Incident Atrial Fibrillation in Fitbit Wearable Users: A Longitudinal Analysis of the All of Us Research Program (AHA 2026 Abstract 11107)

Venue listed2026Added Nov 15, 2026

  • SS
  • AP
  • RD
  • SH
  • PS

Authors: Srinivasulu Yerukala Sathipati, Apurva Popat, Richard A. Dart, Scott Joseph Hebbring, Param Sharma

Circulation 154 suppl 1, AHA Scientific Sessions 2026, Abstract 11107 (2026-11-02)

TL;DR

43,704 Fitbit-linked All of Us participants (≥ 10 valid nights, no prior AF); median follow-up 3.2 y, 335 incident AF. A crude HRV proxy (day-to-day SD of zone-weighted mean HR) was not associated with AF (HR 1.015/SD, p = 0.30). Each extra hour of deep sleep: HR 0.865 (95% CI 0.776–0.965); each extra hour of REM: HR 0.887 (0.815–0.965), adjusted for 16 covariates. AF cases had less deep (0.77 vs 0.99 h) and REM (1.00 vs 1.32 h) sleep.

  • Form factor: wrist (Fitbit)

Abstract

Background: Consumer wearables offer continuous passive monitoring of sleep and cardiac physiology at population scale, yet their utility for predicting atrial fibrillation (AF) remains undefined.

Scientific Questions/Hypothesis: We hypothesized that Fitbit-derived heart rate variability (HRV) and sleep architecture (deep and REM sleep) independently predict incident AF after adjustment for established clinical risk factors.

Methods: We analyzed 43,704 Fitbit-linked participants in the NIH All of Us Research Program with ≥10 nights of valid wearable data and no prior AF diagnosis. An HRV proxy was derived as the day-to-day standard deviation of zone-weighted estimated mean daily heart rate from Fitbit heart rate zone summaries. Sleep architecture (deep/slow-wave and REM sleep hours per night) was extracted from Fitbit sleep-staging data. Incident AF was ascertained via OMOP common data model ICD-10 code linkage to electronic health records. Cox proportional hazards models were adjusted for 16 covariates: age, sex, ethnicity, hypertension, diabetes, chronic kidney disease (CKD), heart failure (HF), obesity, alcohol use disorder, daily step count, hyperthyroidism, valvular heart disease, and COPD. Discriminative performance was quantified by 5-fold cross-validated AUC.

Results: Over a median follow-up of 3.2 years, 335 participants developed incident AF. The wearable HRV proxy was not independently associated with incident AF (HR 1.015 per SD, 95% CI 0.987-1.044, p=0.304). Sleep architecture, however, was a robust independent predictor after full adjustment: each additional hour of deep sleep was associated with a 13.5% reduction in AF risk (HR 0.865, 95% CI 0.776-0.965, p=0.009), and each additional hour of REM sleep with an 11.3% reduction (HR 0.887, 95% CI 0.815-0.965, p=0.005). AF cases had significantly less deep sleep (0.77 vs. 0.99 hrs/night) and REM sleep (1.00 vs. 1.32 hrs/night) than controls (both p Conclusions: In this large wearable-linked cohort, reduced deep and REM sleep — but not a zone-level HRV proxy — independently predicted incident AF after adjustment for all established risk factors. COPD was identified as an additional independent predictor. These findings support passive wearable sleep staging as a novel, scalable AF risk signal. Refinement of wearable HRV metrics beyond heart rate zone aggregates may be required to capture arrhythmia-relevant autonomic signals.

Verbatim from the publisher's DOI record (opens in a new tab) · Abstract as deposited by the publisher with Crossref; © the authors/publisher

Results on ChatPPG leaderboards

All leaderboards

No results from this paper on ChatPPG leaderboards yet. Browse the leaderboards.

Tasks & signals

Tasks

Sleep stagingAF / rhythm

Signals

Keywords

FitbitAll of UsSleep ArchitectureIncident AFEpidemiologyHRV ProxyConference Abstract

How it works

How Method works: Setup → Method → TargetSetup: Signals (PPG (Fitbit summaries), ACC); Form factor (wrist (Fitbit)). Method: Method. Target: Sleep staging; AF / rhythm.SetupSignals: PPG (Fitbit summaries), ACCSignalsPPG (Fitbit summaries), ACCForm factor: wrist (Fitbit)Form factorwrist (Fitbit)MethodMethodTargetSleep stagingSleep stagingAF / rhythmAF / rhythm

Cite this paper

Cite this paper with the fields recorded here. Check it against the publisher before submitting.

Sathipati, S. Y., Popat, A., Dart, R. A., Hebbring, S. J., & Sharma, P. (2026). Sleep Architecture Independently Predicts Incident Atrial Fibrillation in Fitbit Wearable Users: A Longitudinal Analysis of the All of Us Research Program (AHA 2026 Abstract 11107). Circulation 154 suppl 1, AHA Scientific Sessions 2026, Abstract 11107. https://doi.org/10.1161/circ.154.suppl_1.11107
Sathipati, Srinivasulu Yerukala, Apurva Popat, Richard A. Dart, Scott Joseph Hebbring, and Param Sharma. 2026. "Sleep Architecture Independently Predicts Incident Atrial Fibrillation in Fitbit Wearable Users: A Longitudinal Analysis of the All of Us Research Program (AHA 2026 Abstract 11107)." Circulation 154 suppl 1, AHA Scientific Sessions 2026, Abstract 11107. https://doi.org/10.1161/circ.154.suppl_1.11107.
Sathipati, S. Y., Popat, A., Dart, R. A., Hebbring, S. J. and Sharma, P. (2026) 'Sleep Architecture Independently Predicts Incident Atrial Fibrillation in Fitbit Wearable Users: A Longitudinal Analysis of the All of Us Research Program (AHA 2026 Abstract 11107)', Circulation 154 suppl 1, AHA Scientific Sessions 2026, Abstract 11107. doi: 10.1161/circ.154.suppl_1.11107.
Sathipati, Srinivasulu Yerukala, et al. "Sleep Architecture Independently Predicts Incident Atrial Fibrillation in Fitbit Wearable Users: A Longitudinal Analysis of the All of Us Research Program (AHA 2026 Abstract 11107)." Circulation 154 suppl 1, AHA Scientific Sessions 2026, Abstract 11107, 2026. https://doi.org/10.1161/circ.154.suppl_1.11107.
S. Y. Sathipati, A. Popat, R. A. Dart, S. J. Hebbring, and P. Sharma, "Sleep Architecture Independently Predicts Incident Atrial Fibrillation in Fitbit Wearable Users: A Longitudinal Analysis of the All of Us Research Program (AHA 2026 Abstract 11107)," Circulation 154 suppl 1, AHA Scientific Sessions 2026, Abstract 11107, 2026, doi: 10.1161/circ.154.suppl_1.11107.
@article{sathipati2026sleep,
  title         = {{Sleep Architecture Independently Predicts Incident Atrial Fibrillation in Fitbit Wearable Users: A Longitudinal Analysis of the All of Us Research Program (AHA 2026 Abstract 11107)}},
  author        = {Sathipati, Srinivasulu Yerukala and Popat, Apurva and Dart, Richard A. and Hebbring, Scott Joseph and Sharma, Param},
  year          = {2026},
  journal       = {Circulation 154 suppl 1, AHA Scientific Sessions 2026, Abstract 11107},
  doi           = {10.1161/circ.154.suppl_1.11107}
}

Added to the ChatPPG research hub