Icon Legend

This session is not in your schedule.

This session is in your schedule. Click again to remove it.

Presentation Icons

Additional registration fee required

Faculty have requested this content not be shared outside of the session

CME Credit Offered

Abstract Award

Recording available 2/16-5/2

AIUM Credit

Foundation Awardee

Poster Icons

Abstract Award

Foundation Awardee

Poster Session 4

Category: Labor

Poster Session 4

(1135) Identifying Prolonged Second Stage Labor Using Machine Learning Techniques

Thursday, February 12, 2026

3:30 PM - 5:00 PM

Coauthor(s)

SG

Stephanie Guang, MD

Columbia University
New York, New York, United States

Submitting Author and Presenting Author(s)

Katherine L. Grantz, MD, MSCR

Senior Investigator
Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health
Bethesda, Maryland, United States

Coauthor(s)

MB

Mathew Bruckner, MS

The Prospective Group, Inc.
Fairfax, Virginia, United States
AN

Advait Nene, MS

Carnegie Mellon University
Pittsburgh, Pennsylvania, United States
JG

Jessica Gleason, PhD

NICHD, NIH
Bethesda, Maryland, United States
NP

Neil Perkins, PhD

Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health
Bethesda, Maryland, United States
RS

Rajeshwari L. Sundaram, PhD

Senior Investigator
Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health
Bethesda, Maryland, United States

Objective:
Clinical practice of safe duration for the second stage of labor has been evolving. Assessing risk of prolonged second stage is dependent on many dynamic antepartum and intrapartum factors yet prior work used classical regression with a limited number of well-defined variables. We used machine learning models to analyze high-dimensional electronic health record data to predict prolonged second stage of labor.

Study Design:
In the Consortium on Safe Labor, prolonged second stage of labor duration was defined as beyond two hours for nulliparas and one hour for multiparas with an additional hour for an epidural. Sixty-nine risk factors extracted from prenatal and delivery electronic medical records combined with ICD–9 codes were considered. We used logistic regression and random forest to predict prolonged labor, over all deliveries, and stratified by parity. We also studied the predictive ability of models with and without Bishop’s score. Partial dependence plots were used to explore the direction of association of the risk-factors on prolonged second stage labor in this high-dimensional setting.

Results:

Of the 103,415 parturients assessed, 3,519 (3.5%) had prolonged second stage of labor Mean second stage duration was 0.85 hours (SD 1.02). The random forest model had the best performance with area under the curve of 0.80 (95% CI 0.78-0.81) for all deliveries, 0.81 in multiparous and 0.72 in nulliparas.(Table 1) Overall, the most important features selected in best predictive models included parity, station at full dilation, cervical examination data at admission and maternal age.(Figure) Cervical dilation on admission was the predominant prediction factor for multiparas; in contrast, there were multiple factors for nulliparas, with maternal age, admission contractions, epidural and cervical effacement leading.

Conclusion:

These findings demonstrate utility of machine learning in predicting prolonged second stage of labor using data available to the obstetric provider. Machine learning plays a promising role in creating individualized predictions to support clinical risk assessment during labor.