Smartwatches Track Pregnancy Hormones With 93% Accuracy in New Study
Wearable sensors captured resting heart rate patterns that closely mirrored hormonal shifts across pregnancy, flagging adverse outcomes too.
Summary
Researchers at Scripps Research Translational Institute analyzed wearable sensor data from 99 pregnant participants, tracking resting heart rate, steps, and sleep from 3 months pre-pregnancy through 6 months postpartum. They found that resting heart rate changes correlated remarkably closely with known pregnancy hormone fluctuations (R²=0.93). RHR initially dipped in weeks 5–9, rose steadily until 8–9 weeks before delivery, then declined to below pre-pregnancy levels postpartum. Sleep increased in the first trimester but declined thereafter. Importantly, pregnancies with adverse outcomes showed distinct RHR patterns compared to live births, suggesting wearables may one day enable early risk screening for miscarriage or preterm birth without clinical visits.
Detailed Summary
The United States faces a worsening maternal health crisis, with adverse pregnancy outcomes disproportionately affecting non-White women and those in maternity care deserts where access to obstetric services is severely limited. This study explored whether consumer-grade wearable devices could passively capture the profound physiological changes of pregnancy—and whether those signals mirror the well-characterized hormonal shifts that drive them.
Using the PowerMom bilingual digital research platform, investigators recruited participants aged 16 or older who were pregnant or up to 8 weeks postpartum. Of 5,612 enrolled participants, 99 live-birth pregnancies met data-quality thresholds (≥2 valid wearable days per week for ≥50% of pregnancy weeks) for resting heart rate (RHR) analysis, with 54 contributing sleep data. Devices included Apple Watch, Garmin, and Fitbit. To control for inter-individual and inter-device variability, all signals were expressed relative to each participant's pre-pregnancy baseline and aggregated at the population level.
The headline finding was a striking R²=0.93 correlation between the longitudinal RHR trajectory and expected pregnancy hormone profiles. RHR fell modestly in gestational weeks 5–9—consistent with the early vasodilatory effects of progesterone and hCG—then climbed steadily as cardiac output rises through mid-pregnancy, peaking roughly 8–9 weeks before delivery. It then declined toward and below pre-pregnancy levels postpartum before stabilizing near baseline at 6 months. Total sleep time rose in the first trimester but progressively declined through the remainder of pregnancy. Step counts showed activity reductions as pregnancy advanced. These patterns collectively reflect well-known haemodynamic and behavioural adaptations to pregnancy.
An exploratory analysis of 9 participants with adverse outcomes (miscarriage or other non-live-birth endings) revealed RHR trajectories that diverged meaningfully from the live-birth group, hinting that wearables may detect physiological deviations before clinical symptoms manifest. The authors are careful to frame this as a feasibility signal given the small adverse-outcome sample, not a validated predictive tool.
The study's significance lies in demonstrating that unobtrusive, consumer-level technology worn daily can passively reconstruct biologically coherent pregnancy timelines—potentially extending meaningful prenatal monitoring to the 30–50% of birthing individuals in maternity care deserts. Future prospective studies with larger adverse-outcome cohorts are needed to translate these associations into actionable clinical predictions.
Key Findings
- Wearable RHR trajectories correlated with pregnancy hormone fluctuations at R²=0.93 across 99 live-birth pregnancies.
- RHR dipped in weeks 5–9, rose to a peak ~8–9 weeks before delivery, then fell below pre-pregnancy levels postpartum.
- Total sleep time increased in the first trimester but declined progressively through the rest of pregnancy.
- Adverse-outcome pregnancies showed distinct RHR patterns vs. live births, suggesting early detection potential.
- Passive wearable monitoring captured population-level pregnancy physiology without requiring clinical visits.
Methodology
Longitudinal observational cohort study using the PowerMom bilingual mobile research platform; 99 live-birth participants contributed Apple Watch, Garmin, or Fitbit data spanning 3 months pre-pregnancy to 6 months postpartum. Signals were normalized to pre-pregnancy baselines and required ≥2 valid wearable days per week for ≥50% of pregnancy weeks.
Study Limitations
The adverse-outcome group contained only 9 participants, making those comparisons exploratory and underpowered. Pregnancy start was estimated from self-reported due dates rather than confirmed gestational dating, introducing potential timing error. Sleep analysis was restricted to Fitbit users (n=54), limiting generalizability across device types.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.