Machine learning predicts risk of preterm delivery

Graph depicting preterm birth likelihood

Predicting the likelihood of delivery preterm using data collected from a pregnancy mobile app

Published: American College of Obstetricians and Gynecologists (ACOG)
Authors: Bradley, Danielle MS, MPH; Landau, Erin; Wolfberg, Adam MD, MPH; Baron, Alex PhD

Risk for preterm delivery is largely only identified among women with previous preterm deliveries or explicit risk factors, like insufficient cervix. But for the other group of seemingly low risk women who end up delivering preterm, there are very few proven methods for identifying risk in time to intervene.

Ovia Health’s data science team applied machine learning methodologies to determine our ability to detect the likelihood of delivering preterm. We were able to detect 41% of cases with 75% positive predictive value. The results of which were peer-reviewed and presented at the American College of Obstetricians and Gynecologists (ACOG) annual conference.

About Ovia Health

Ovia Health has helped over 14 million women and families. We offer a comprehensive maternity and family benefits solution from preconception and pregnancy through return-to-work and parenthood. 

Ovia Health:

  • Engages women early, and sustains long term daily participation
  • Actively encourages partners to participate in the process
  • Includes predictive coaching and mobile alerts for identified health issues
  • Supports return-to-work with a proven approach
  • Boosts utilization across an employer’s benefit ecosystem
  • Drives behavior change and reduces maternity costs