Wearable-derived Maternal Heart Rate Variability As a Novel Digital Biomarker of Preterm Birth
Overview
Affiliations
Despite considerable health consequences from preterm births, their incidence remains unchanged over recent decades, due partially to limited screening methods and limited use of extant methods. Wearable technology offers a novel, noninvasive, and acceptable way to track vital signs, such as maternal heart rate variability (mHRV). Previous research observed that mHRV declines throughout the first 33 weeks of gestation in term, singleton pregnancies, after which it improves. The aim of this study was to explore whether mHRV inflection is a feature of gestational age or an indication of time to delivery. This retrospective case-control study considered term and preterm deliveries. Remote data collection via non-invasive wearable technology enabled diverse participation with subjects representing 42 US states and 16 countries. Participants (N = 241) were retroactively identified from the WHOOP (Whoop, Inc.) userbase and wore WHOOP straps during singleton pregnancies between March 2021 and October 2022. Mixed effect spline models by gestational age and time until birth were fit for within-person mHRV, grouped into preterm and term births. For term pregnancies, gestational age (Akaike information criterion (AIC) = 26627.6, R2m = 0.0109, R2c = 0.8571) and weeks until birth (AIC = 26616.3, R2m = 0.0112, R2c = 0.8576) were representative of mHRV trends, with significantly stronger fit for weeks until birth (relative log-likelihood ratio = 279.5). For preterm pregnancies, gestational age (AIC = 1861.9, R2m = 0.0016, R2c = 0.8582) and time until birth (AIC = 1848.0, R2m = 0.0100, R2c = 0.8676) were representative of mHRV trends, with significantly stronger fit for weeks until birth (relative log-likelihood ratio = 859.4). This study suggests that wearable technology, such as the WHOOP strap, may provide a digital biomarker for preterm delivery by screening for changes in nighttime mHRV throughout pregnancy that could in turn alert to the need for further evaluation and intervention.
Deep learning model using continuous skin temperature data predicts labor onset.
Basavaraj C, Grant A, Aras S, Erickson E BMC Pregnancy Childbirth. 2024; 24(1):777.
PMID: 39587525 PMC: 11587739. DOI: 10.1186/s12884-024-06862-9.
Preterm birth risk stratification through longitudinal heart rate and HRV monitoring in daily life.
Feli M, Azimi I, Sarhaddi F, Sharifi-Heris Z, Niela-Vilen H, Liljeberg P Sci Rep. 2024; 14(1):19896.
PMID: 39191907 PMC: 11349982. DOI: 10.1038/s41598-024-70773-0.
Sharifi-Heris Z, Yang Z, Rahmani A, Fortier M, Sharifiheris H, Bender M Front Physiol. 2023; 14:1293946.
PMID: 38074317 PMC: 10702512. DOI: 10.3389/fphys.2023.1293946.
Erickson E, Gotlieb N, Pereira L, Myatt L, Mosquera-Lopez C, Jacobs P NPJ Digit Med. 2023; 6(1):153.
PMID: 37598232 PMC: 10439919. DOI: 10.1038/s41746-023-00902-y.