Association between wearable sensor signals and expected hormonal changes in pregnancy.

Background: The rising maternal health crisis in the United States necessitates innovative approaches to pregnancy monitoring. This observational cohort study aimed to assess whether wearable sensors can effectively track physiological and behavioural changes during pregnancy and examine their associations with pregnancy-related hormonal fluctuations.

Methods: This longitudinal cohort study recruited participants via a bilingual mobile research platform. Eligible participants were aged ≥16 years, pregnant or within eight weeks postpartum, residing in the United States, and consented to share data collected by their own wearable device (Apple, Garmin, or Fitbit). Self-reported survey responses and wearable sensor data were collected from 99 participants who experienced a live birth pregnancy, from three months pre-pregnancy to six months postpartum. Key outcomes included changes in resting heart rate (RHR), physical activity, and sleep patterns, analysed in relation to expected hormonal fluctuations during pregnancy.

Findings: In live birth pregnancies, RHR initially decreased between weeks 5-9, followed by a steady increase until 8-9 weeks before delivery, then declined until birth, dropping below pre-pregnancy levels postpartum before stabilising at six months. Total sleep time increased in the first trimester but decreased throughout the remainder of pregnancy. A strong correlation was observed between RHR fluctuations and pregnancy-induced hormonal changes (R = 0.93). Pregnancies ending in adverse outcomes displayed distinct RHR patterns compared to live birth pregnancies.

Interpretation: These findings suggest that wearable sensors provide a non-invasive method to monitor pregnancy-related physiological and behavioural changes, which align with hormonal shifts. This study provides population-level insights in live birth pregnancies, and an exploratory analysis of adverse outcomes reflecting the feasibility of recruiting and capturing physiological signals also in these cases. This approach may enable early risk assessment for adverse pregnancy outcomes, including miscarriage and preterm birth.

Funding: This work was supported by the National Center for Advancing Translational Sciences (UM1TR004407) and from the Patrick J. McGovern Foundation.