The impact of healthy pregnancy on features of heart rate variability and pulse wave morphology derived from wrist‑worn photoplethysmography

Pregnancy induces significant, continuous changes in the maternal autonomic nervous system to maintain homeostasis and support fetal development. Dysfunctional autonomic regulation is linked to adverse pregnancy outcomes, such as hypertensive disorders. While researchers often use heart rate variability (HRV) from wearable sensors to track these changes, it remains unclear which specific features are most diagnostic. This study evaluates a broad set of HRV and pulse wave morphological features derived from wrist-worn photoplethysmography (PPG) to determine which are best at identifying the physiological signatures of a healthy pregnancy.

The researchers performed a retrospective analysis of nighttime PPG data collected from 36 healthy pregnant women (second and third trimesters) and 36 healthy non-pregnant women. They processed the PPG signals to extract a wide array of HRV metrics—including time-domain, frequency-domain, non-linear, phase-rectified signal averaging (PRSA), and heart rate fragmentation (HRF) features—alongside a comprehensive set of morphological features describing the PPG waveform. Using logistic regression and stepwise forward feature elimination, they ranked the predictive power of these features to classify pregnancy status. They also explored the impact of stratifying data by sleep stages, specifically analyzing whether features from deep sleep (N3) offered improved discrimination.

Morphological features consistently outperformed HRV features, achieving higher classification accuracy (AUROC of 0.825 vs. 0.74). The single most predictive feature was systolic pulse wave duration (SPD), followed by mean heart rate. The authors also found that analyzing data specifically from deep sleep (N3) enhanced the ability to distinguish between the two groups, likely because deep sleep provides a more stable, parasympathetically dominated environment that minimizes motion artifacts and physiological arousals. While the study suggests that morphological analysis is a powerful addition to maternal health monitoring, the authors note that further research is required to fully characterize the physiological underpinnings of these pulse wave changes.