“`html
Complications arising from preterm births are the leading reason for mortality among children below 5 years of age, and nearly 75% could be avoided through interventions, as noted by the World Health Organization. Although the factors causing preterm births remain largely elusive, a multidisciplinary research group at Washington University in St. Louis has discovered that fluctuations in sleep patterns among pregnant individuals can effectively anticipate preterm births.
Ben Warner, a doctorate candidate at the McKelvey School of Engineering, and Peinan Zhao, an assistant professor in obstetrics and gynecology at WashU Medicine, utilized machine learning models to examine sleep data from expectant participants. While sleep disruption is recognized as a predictor of preterm birth, which occurs prior to 37 weeks’ gestation, the underlying causes have remained unclear as the data relied on self-reporting from patients.
Warner analyzed data from 665 patients during their first two trimesters who had a documented delivery date. These patients were part of a 2014 cohort study executed at Washington University in St. Louis and BJC HealthCare. Approximately 14% of this group experienced a preterm birth. Warner is mentored by Chenyang Lu, the Fullgraf Professor in the Computer Science & Engineering Department at McKelvey Engineering; Zhao; and Sarah England, the Alan A. and Edith L. Wolff Professor and vice chair for research in obstetrics and gynecology at WashU Medicine.
The patients utilized a clinically validated wristwatch, referred to as an actigraph, which tracked body movements over roughly two-week intervals. This data enabled the team to identify daily trends in sleep duration, bedtimes, wake times, movement during sleep, and various other factors. Additionally, patients completed surveys detailing their sleep habits. Warner and Zhao integrated these two data sources and employed machine-learning models to evaluate the influence of sleep patterns on preterm births.
“We discovered that sleep metrics are reasonably predictive of preterm birth,” Warner remarked. “Variability in sleep patterns typically serves as a more robust predictor of preterm birth compared to average sleep metrics, and maintaining a consistent sleep routine is more crucial than simply achieving good sleep on average.”
Lu’s lab has examined data from wearable technologies for a range of issues, including contact tracing for COVID-19, forecasting surgical outcomes, and identifying symptoms of depression and anxiety.
“Raw data from wearables can be quite chaotic, but by employing a comprehensive mix of statistical techniques, AI (artificial intelligence), and clinical expertise, researchers can derive vital clinical insights,” stated Lu, who also directs the university’s AI for Health Institute. “Subsequently, AI scientists and clinicians collaborate to distill insights from these complex real-world data sets and extract meaningful information.”
Zhao indicated that their models were deliberately straightforward, focusing on clinically relevant associations within the results.
“Our model demonstrates that a machine-learning approach is superior to a purely statistical model,” Zhao asserted, who is also part of the Center for Reproductive Health Sciences. “We can analyze how much each variable contributes to the predictive model and its impact on the final outcome. From this, a potential intervention could be to support a more regular sleep schedule.”
While many pregnant individuals report sleep disturbances, these generally manifest during the third trimester, according to England. The team concentrated on identifying signals before women reached the 20-week mark in their pregnancies.
“There is currently no intervention since we can’t predict who will have a preterm birth,” England noted, directing the Center for Reproductive Health Sciences. “We hope this research will significantly enhance our ability to identify women at higher risk.”
Looking forward, the team intends to validate their findings in different populations at various academic medical centers.
“This study underscores the collaboration between engineering and obstetrics and gynecology,” England remarked. “Many do not recognize the inherent connections between engineering and reproductive health, and this serves as yet another example of a fruitful partnership between engineers and researchers in our field.”
Warner BC, Zhao P, Herzog ED, Frolova AI, England SK, Lu C. Validation of sleep-based actigraphy machine learning models for prediction of preterm birth. npj Women’s Health. June 20, 2025, DOI: 10.1038/s44294-025-00082-y.
Funding for this study was provided by the March of Dimes Foundation, the Fullgraf Foundation, St. Louis Children’s Hospital, Barnes-Jewish Hospital, the Department of Obstetrics and Gynecology at Washington University School of Medicine in St. Louis, Washington University AI for Health Institute, and the James McKelvey School of Engineering at Washington University in St. Louis.
The code utilized in this research is accessible at https://github.com/bcwarner/mod-actigraphy-clf
Initially published on the McKelvey School of Engineering website
The article Sleep data from wearable device may help predict preterm birth first appeared on The Source.
“`