Deep Time-to-Event Models for Intrapartum Fetal Monitoring
- 1Boston Children's Hospital, Harvard Medical School
Abstract
Timely detection of fetal distress during labor is a central pursuit of modern obstetric care. The primary measure used in this regard is fetal heart rate and contraction monitoring (echocardiotocography or CTG), which provides a continuous signal during labor. The vast majority of machine learning applications to CTG assume access to a 30 to 60 minute window just prior to delivery, which makes an untenable assumption that the timing of delivery is known. In this work, we reframe the problem as a time-to-event prediction task. We develop deep time-to-event models to predict the joint probability of imminent delivery and fetal acidosis, a biomarker of fetal hypoxia collected from umbilical cord gas after delivery. We develop a framework for this problem, drawing on deep discrete-time survival methods and marked point processes, adapting each to this particular setting (i.e., continuous physiological signals, fully observed events, partially observed labels). One approach dubbed Marked DeepHit accurately estimates delivery within one hour at multiple elapsed-time landmarks, with test-set AUROC of 0.837 [95% CI 0.692-0.970] and 0.935 [0.858-0.964] at 6 and 12 hours into labor, respectively. We find that these models also achieve competitive performance near delivery time compared to prior models trained specifically on the end of tracing, and outperform other clinical feature-based proposals, both on internal and external validation at a second site. These results give a proof-of-principle to support future prospective deployment of AI-EFM during labor.
Related
- Introducing the Moyo device for intrapartum fetal heart rate monitoring among women with high-risk pregnancies in a low-resource setting: a quasi-experimental study
- Enhancing prenatal diagnostic workflow: a deep learning-based system for multi-class fetal ultrasound classification across all gestational stages
- Simultaneous acquisition of maternal and dual-fetal electrocardiography with maternal haemodynamic monitoring in anaesthetized pregnant rats.
- Time-aware machine learning enables early intrapartum identification of fetuses at risk of hypoxic-ischemic encephalopathy from cardiotocography
- Anticholinergic Co-Medication Burden Does Not Predict Cardiac Toxicity in Adolescent Diphenhydramine Exposure: A Pharmacovigilance Study of the FDA Adverse Event Monitoring System
BibTeX
@misc{aldeia2026deep,
title = {{Deep Time-to-Event Models for Intrapartum Fetal Monitoring}},
author = {G. S. Imai Aldeia and H. Coggan and Y. Yang and L. Levine and J. A. McCoy and W. La Cava},
howpublished = {medRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.21.26363416},
url = {https://www.medrxiv.org/content/10.64898/2026.09.21.26363416v1},
note = {CC BY-NC},
}
RIS
TY - UNPB
TI - Deep Time-to-Event Models for Intrapartum Fetal Monitoring
AU - G. S. Imai Aldeia
AU - H. Coggan
AU - Y. Yang
AU - L. Levine
AU - J. A. McCoy
AU - W. La Cava
PY - 2026
DA - 2026/09/21
PB - medRxiv
DO - 10.64898/2026.09.21.26363416
UR - https://www.medrxiv.org/content/10.64898/2026.09.21.26363416v1
AB - Timely detection of fetal distress during labor is a central pursuit of modern obstetric care. The primary measure used in this regard is fetal heart rate and contraction monitoring (echocardiotocography or CTG), which provides a continuous signal during labor. The vast majority of machine learning applications to CTG assume access to a 30 to 60 minute window just prior to delivery, which makes an untenable assumption that the timing of delivery is known. In this work, we reframe the problem as a time-to-event prediction task. We develop deep time-to-event models to predict the joint probability of imminent delivery and fetal acidosis, a biomarker of fetal hypoxia collected from umbilical cord gas after delivery. We develop a framework for this problem, drawing on deep discrete-time survival methods and marked point processes, adapting each to this particular setting (i.e., continuous physiological signals, fully observed events, partially observed labels). One approach dubbed Marked DeepHit accurately estimates delivery within one hour at multiple elapsed-time landmarks, with test-set AUROC of 0.837 [95% CI 0.692-0.970] and 0.935 [0.858-0.964] at 6 and 12 hours into labor, respectively. We find that these models also achieve competitive performance near delivery time compared to prior models trained specifically on the end of tracing, and outperform other clinical feature-based proposals, both on internal and external validation at a second site. These results give a proof-of-principle to support future prospective deployment of AI-EFM during labor.
ER -