Mixed-Effects Machine Learning improved prediction of Obstetric Anal Sphincter Injury: a nationwide multicenter cohort study.
- 1Amsterdam UMC Location AMC: Amsterdam UMC Locatie AMC
Abstract
Existing prediction models for obstetric anal sphincter injuries (OASIS) often rely on traditional regression, lack external validation, and, when fitted on multicenter data, fail to account for hospital-level clustering. We aimed to develop and temporally validate mixed-effects machine learning models for OASIS and compare them with traditional regression models. A secondary aim was to evaluate whether macrosomia ([≥]4000 g) could substitute for continuous birth weight. We conducted a retrospective nationwide cohort study of singleton, term ([≥]37 weeks), cephalic vaginal deliveries in the Dutch Perinatal Registry. Data from 2016 to 2019 were used for model development with three-fold cross-validation; data from 2020 were reserved for temporal validation. Candidate predictors included 18 demographic, obstetric, and neonatal factors. We developed six models: mixed-effects gradient boosting (MEGB), mixed-effects LASSO (ME-LASSO), and backward stepwise logistic regression, each with either birth weight or macrosomia. Discrimination was assessed by the area under the receiver operating characteristic curve (AUC), calibration by calibration plots, and prediction accuracy by the Brier score. Among 614,255 deliveries, OASIS occurred in 2.6%. In temporal validation including macrosomia, MEGB achieved the highest discrimination (AUC 0.727; 95% CI 0.719-0.735), significantly outperforming ME-LASSO (AUC 0.719; 95% CI 0.710-0.727; P<.0001) and stepwise logistic regression (AUC 0.714; 95% CI 0.706-0.722; P<.0001). Including birth weight marginally improved performance, with macrosomia serving as an effective proxy. Consistently important predictors were nulliparity, prolonged second stage, mediolateral episiotomy, advanced maternal age, operative vaginal delivery, macrosomia, and previous cesarean delivery. Uncalibrated MEGB showed poorer calibration, but isotonic recalibration substantially improved it. After calibration, all models showed excellent agreement between predicted and observed risks. MEGB provided improved discrimination and comparable calibration relative to ME-LASSO and stepwise logistic regression. Replacing continuous birth weight with a binary macrosomia indicator resulted in minimal performance loss, suggesting it may suffice for clinical risk prediction across all three modeling approaches.
Author summaryObstetric Anal Sphincter Injuries (OASIS) after giving birth, can lead to long-term issues such as incontinence, pelvic floor dysfunction, chronic pain, and sexual dysfunction. Identifying women at increased risk before birth could support counselling and help clinicians take preventive measures. Although several prediction tools have been developed, many do not account for possible differences between hospitals, where clinical practices and rates of OASIS may vary.
Using nationwide data from more than 600,000 births in the Netherlands, we developed prediction models for OASIS that also accounted for differences between hospitals. We used machine-learning approaches and compared their performance with that of a traditional regression model. The machine-learning models performed slightly better in predicting OASIS. We also found that classifying babies as large (birth weight above 4,000 g) performed almost as well as using exact birth weight. This is relevant because fetal weight can only be estimated before birth using ultrasound. After further external validation, these models may support earlier and more individualized assessment of OASIS risk.
Related
- Real-World Adaptation and External Validation of Machine-Learning Impedance Spectroscopy for Immediate Postpartum Detection of Obstetric Anal Sphincter Injuries
- Improving Immediate Postpartum Detection of Obstetric Anal Sphincter Injury: Added Diagnostic Value of Impedance Spectroscopy Following Clinical Assessment
- Lived experiences of obstetric anal sphincter injury among women from ethnically diverse backgrounds in the UK: a qualitative study.
- Perineal Body Reconstruction for Anal Sphincter Insufficiency Following Obstetric Perineal Body Injury: Morphology-based Classification and Type-stratified Outcomes in 42 Cases.
- Intrapartum rectovaginal distance as a predictor of obstetric anal sphincter injury: a prospective cohort study.
BibTeX
@misc{mohamed2026mixedeffects,
title = {{Mixed-Effects Machine Learning improved prediction of Obstetric Anal Sphincter Injury: a nationwide multicenter cohort study.}},
author = {A. Chahid Mohamed and A. C. Ravelli and J. de Leeuw and J. van Bavel and A. Abu-Hanna},
howpublished = {medRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.29.26364280},
url = {https://www.medrxiv.org/content/10.64898/2026.09.29.26364280v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Mixed-Effects Machine Learning improved prediction of Obstetric Anal Sphincter Injury: a nationwide multicenter cohort study.
AU - A. Chahid Mohamed
AU - A. C. Ravelli
AU - J. de Leeuw
AU - J. van Bavel
AU - A. Abu-Hanna
PY - 2026
DA - 2026/09/29
PB - medRxiv
DO - 10.64898/2026.09.29.26364280
UR - https://www.medrxiv.org/content/10.64898/2026.09.29.26364280v1
AB - Existing prediction models for obstetric anal sphincter injuries (OASIS) often rely on traditional regression, lack external validation, and, when fitted on multicenter data, fail to account for hospital-level clustering. We aimed to develop and temporally validate mixed-effects machine learning models for OASIS and compare them with traditional regression models. A secondary aim was to evaluate whether macrosomia ([≥]4000 g) could substitute for continuous birth weight. We conducted a retrospective nationwide cohort study of singleton, term ([≥]37 weeks), cephalic vaginal deliveries in the Dutch Perinatal Registry. Data from 2016 to 2019 were used for model development with three-fold cross-validation; data from 2020 were reserved for temporal validation. Candidate predictors included 18 demographic, obstetric, and neonatal factors. We developed six models: mixed-effects gradient boosting (MEGB), mixed-effects LASSO (ME-LASSO), and backward stepwise logistic regression, each with either birth weight or macrosomia. Discrimination was assessed by the area under the receiver operating characteristic curve (AUC), calibration by calibration plots, and prediction accuracy by the Brier score. Among 614,255 deliveries, OASIS occurred in 2.6%. In temporal validation including macrosomia, MEGB achieved the highest discrimination (AUC 0.727; 95% CI 0.719-0.735), significantly outperforming ME-LASSO (AUC 0.719; 95% CI 0.710-0.727; P<.0001) and stepwise logistic regression (AUC 0.714; 95% CI 0.706-0.722; P<.0001). Including birth weight marginally improved performance, with macrosomia serving as an effective proxy. Consistently important predictors were nulliparity, prolonged second stage, mediolateral episiotomy, advanced maternal age, operative vaginal delivery, macrosomia, and previous cesarean delivery. Uncalibrated MEGB showed poorer calibration, but isotonic recalibration substantially improved it. After calibration, all models showed excellent agreement between predicted and observed risks. MEGB provided improved discrimination and comparable calibration relative to ME-LASSO and stepwise logistic regression. Replacing continuous birth weight with a binary macrosomia indicator resulted in minimal performance loss, suggesting it may suffice for clinical risk prediction across all three modeling approaches. Author summaryObstetric Anal Sphincter Injuries (OASIS) after giving birth, can lead to long-term issues such as incontinence, pelvic floor dysfunction, chronic pain, and sexual dysfunction. Identifying women at increased risk before birth could support counselling and help clinicians take preventive measures. Although several prediction tools have been developed, many do not account for possible differences between hospitals, where clinical practices and rates of OASIS may vary. Using nationwide data from more than 600,000 births in the Netherlands, we developed prediction models for OASIS that also accounted for differences between hospitals. We used machine-learning approaches and compared their performance with that of a traditional regression model. The machine-learning models performed slightly better in predicting OASIS. We also found that classifying babies as large (birth weight above 4,000 g) performed almost as well as using exact birth weight. This is relevant because fetal weight can only be estimated before birth using ultrasound. After further external validation, these models may support earlier and more individualized assessment of OASIS risk.
ER -