PDF icon Download PDF
Am J Perinatol 2025; 42(03): 281-292
DOI: 10.1055/a-2405-3459
SMFM Fellowship Series Article

Machine Learning for the Prediction of Surgical Morbidity in Placenta Accreta Spectrum

Itamar D. Futterman 
1   Division of Maternal Fetal Medicine, Department of Obstetrics and Gynecology, Maimonides Medical Center, Brooklyn, New York
2   Division of Complex Obstetrical Surgery, Department of Obstetrics and Gynecology, Maimonides Medical Center, Brooklyn, New York
,
Olivia Sher
1   Division of Maternal Fetal Medicine, Department of Obstetrics and Gynecology, Maimonides Medical Center, Brooklyn, New York
,
Chaskin Saroff
3   BioticAI Inc., San Francisco, California
,
Alexa Cohen
4   Division of Fetal Medicine and Ultrasound, Obstetrics, Gynecology and Women's Health, Montefiore Medical Center, Albert Einstein College of Medicine, Bronx, New York
,
Georgios Doulaveris
4   Division of Fetal Medicine and Ultrasound, Obstetrics, Gynecology and Women's Health, Montefiore Medical Center, Albert Einstein College of Medicine, Bronx, New York
,
Pe'er Dar
4   Division of Fetal Medicine and Ultrasound, Obstetrics, Gynecology and Women's Health, Montefiore Medical Center, Albert Einstein College of Medicine, Bronx, New York
,
Myah M. Griffin
5   Division of Maternal Fetal Medicine, Department of Obstetrics and Gynecology, NYU Langone Medical Center, New York, New York
,
Meghana Limaye 
5   Division of Maternal Fetal Medicine, Department of Obstetrics and Gynecology, NYU Langone Medical Center, New York, New York
,
Thomas Owens
6   Division of Maternal Fetal Medicine, Mount Sinai West, Icahn School of Medicine at Mount Sinai, New York, New York
,
Lois Brustman
6   Division of Maternal Fetal Medicine, Mount Sinai West, Icahn School of Medicine at Mount Sinai, New York, New York
,
Henri Rosenberg
7   Division of Maternal Fetal Medicine, Department of Obstetrics, Gynecology and Reproductive Science, Icahn School of Medicine at Mount Sinai, New York, New York
,
Rebecca Jessel
7   Division of Maternal Fetal Medicine, Department of Obstetrics, Gynecology and Reproductive Science, Icahn School of Medicine at Mount Sinai, New York, New York
,
Scott Chudnoff
1   Division of Maternal Fetal Medicine, Department of Obstetrics and Gynecology, Maimonides Medical Center, Brooklyn, New York
,
Shoshana Haberman
1   Division of Maternal Fetal Medicine, Department of Obstetrics and Gynecology, Maimonides Medical Center, Brooklyn, New York
› Author Affiliations
Funding None.
› Further Information
Also available ateRefHebammen
   Permissions and Reprints
Preview

Abstract

Objective We sought to create a machine learning (ML) model to identify variables that would aid in the prediction of surgical morbidity in cases of placenta accreta spectrum (PAS).

Study Design A multicenter analysis including all cases of PAS identified by pathology specimen confirmation, across five tertiary care perinatal centers in New York City from 2013 to 2022. We developed models to predict operative morbidity using 213 variables including demographics, obstetrical information, and limited prenatal imaging findings detailing placental location. Our primary outcome was prediction of a surgical morbidity composite defined as including any of the following: blood loss (>1,500 mL), transfusion, intensive care unit admission, vasopressor use, mechanical ventilation/intubation, and organ injury. A nested, stratified, cross-validation approach was used to tune model hyperparameters and estimate generalizability. Gradient boosted tree classifier models incorporated preprocessing steps of standard scaling for numerical variables and one-hot encoding for categorical variables. Model performance was evaluated using area under the receiver operating characteristic curve (AUC), positive and negative predictive values (PPV, NPV), and F1 score. Variable importance ranking was also determined.

Results Among 401 PAS cases, 326 (81%) underwent hysterectomy. Of the 401 cases of PAS, 309 (77%) had at least one event defined as surgical morbidity. Our predictive model had an AUC of 0.79 (95% confidence interval: 0.69, 0.89), PPV 0.79, NPV 0.76, and F1 score of 0.88. The variables most predictive of surgical morbidity were completion of a hysterectomy, prepregnancy body mass index (BMI), absence of a second trimester ultrasound, socioeconomic status zip code, BMI at delivery, number of prenatal visits, and delivery time of day.

Conclusion By identifying social and obstetrical characteristics that increase patients' risk, ML models are useful in predicting PAS-related surgical morbidity. Utilizing ML could serve as a foundation for risk and complexity stratification in cases of PAS to optimize surgical planning.

Key Points

  • ML models are useful models are useful in predicting PAS-related surgical morbidity.

  • Optimal management for PAS remains unclear.

  • Utilizing ML can serve as a foundation for risk and complexity stratification in cases of PAS.

Keywords

placenta accreta spectrum - surgical morbidity - machine learning

Note

This work was presented as poster at the Annual Pregnancy Meeting, Society for Maternal Fetal Medicine, National Harbor, MD, February 10–14, 2024.


Precis

Machine learning models are useful in predicting PAS-related surgical morbidity.


Supplementary Material



Publication History

Received: 04 June 2024

Accepted: 25 August 2024

Article published online:
17 September 2024

© 2024. Thieme. All rights reserved.

Thieme Medical Publishers, Inc.
333 Seventh Avenue, 18th Floor, New York, NY 10001, USA