Prediction of Hypertension and Diabetes in Twin Pregnancy Using Machine Learning Model Based on Characteristics at First Prenatal Visit - National Registry Study

Q: What is the study’s main objective?

To develop prediction models for hypertensive disorders of pregnancy (HDP) and gestational diabetes mellitus (GDM) in twin pregnancies using first-prenatal-visit characteristics.

Q: How were the prediction models built and evaluated?

Associations were tested using univariable and multivariable logistic regression. Prediction models used generalized linear models (logit link) and an XGboost machine-learning approach, evaluated with repeated 2-fold cross-validation and AUC.

Q: What did the results show about model performance?

Machine learning and logistic regression had mostly similar predictive performance with negligible differences across tested domains. Models showed modest accuracy and were well calibrated without poor fit.

Aug 5, 2026

Upload by Aurora

Objective: develop a prediction model for hypertensive disorders of pregnancy (HDP) and gestational diabetes mellitus (GDM) in twin pregnancy using characteristics measured at the first prenatal visit. Methods: cross-sectional analysis of national live-birth data in the USA from 2016–2021; candidate variables were evaluated by univariable and multivariable logistic regression, and machine-learning models were trained using generalized linear models and XGboost, with repeated 2-fold cross-validation and AUC assessment. Results: 707,198 twins for HDP and 723,882 for GDM; incidence increased from 2016 to 2021. Performance of machine learning and logistic regression was mostly similar, with modest AUC and adequate calibration. Conclusions: predictive accuracy was comparable across approaches, with modest but well-calibrated models and no evidence of poor fit.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages11

Reading Time28 min

AccessFree to download

Abstract

Objective

Methods

Results

Conclusions

Introduction

Ultrasound Obstet Gynecol 2025; 65: 613–623 Published online in Wiley Online Library ([wileyonlinelibrary.com](wileyonlinelibrary.com)). DOI: 10.1002/uog.27710. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. Prediction of hypertension and diabetes in twin pregnancy using machine learning model based on characteristics at ﬁrst prenatal visit: national registry study H. J. MUSTAFA1,2, E. KALAFAT3, S. PRASAD4, M.-H. HEYDARI5 , R. N. NUNGE6 and A. KHALIL4,7,8 1 Department of Obstetrics and Gynecology, Division of Maternal – Fetal Medicine, Indiana University School of Medicine, Indianapolis, IN, USA; 2 Riley Children and Indiana University Health Fetal Center, Indianapolis, IN, USA; 3 Department of Obstetrics and Gynecology, Koc University School of Medicine, Istanbul, Turkey; 4 Fetal Medicine Unit, St George’s University Hospitals NHS Foundation Trust, University of London, London, UK; 5Non-Communicable Diseases Research Center, Endocrinology and Metabolism Population Sciences Institute, Tehran University of Medical Sciences, Tehran, Iran; 6 Indiana University School of Medicine, Indianapolis, IN, USA; 7 Vascular Biology Research Centre, Molecular and Clinical Sciences Research Institute, St George’s University of London, London, UK; 8 Fetal Medicine Unit, Liverpool Women’s Hospital, Liverpool, UK KEYWORDS: artiﬁcial intelligence; cross-sectional study; gestational diabetes mellitus; gestational hypertension; machine learning; pre-eclampsia; twin pregnancy | ABSTRACT
Objective To develop a prediction model for hypertensive disorders of pregnancy (HDP) and gestational diabetes mellitus (GDM) in twin pregnancy using characteristics obtained at the ﬁrst prenatal visit.
Methods This was a cross-sectional study using national live-birth data in the USA between 2016 and 2021 . The association of all prenatal candidate variables with HDP and GDM was tested on univariable and multivariable logistic regression analyses. Prediction models were built with generalized linear models using the logit link function and classiﬁcation and regression tree (XGboost) machine learning algorithm. Performance was assessed with repeated 2-fold cross-validation and the area under the receiver-operating-characteristics curve (AUC) was calculated. A P value < 0.001 was considered statistically signiﬁcant.
Results A total of 707 198 twin pregnancies were included in the HDP analysis and 723 882 twin pregnancies were included in the GDM analysis. The incidence of HDP and GDM increased signiﬁcantly from 12.6% and 8. 1%, respectively, in 2016 to 16.0% and 10. 7%, respectively, in 2021 . Factors associated with increased odds of HDP in twin pregnancy were maternal age < 20 years or ≥ 35 years, infertility treatment, prepregnancy diabetes mellitus, non-Hispanic | Black race, overweight prepregnancy BMI, prepregnancy obesity and Medicaid as the payment source for delivery (P < 0.001 for all). Obesity Class II and III more than doubled the odds of HDP. Factors associated with increased odds of GDM in twin pregnancy were maternal age ≤ 24 years or ≥ 30 years, infertility treatment, prepregnancy hypertension, non-Hispanic Asian race, maternal birthplace outside the USA and prepregnancy obesity (P < 0.001 for all). Maternal age ≥ 30 years, non-Hispanic Asian race and obesity Class I, II and III more than doubled the odds of GDM. For both HDP and GDM, the performances of the machine learning model and logistic regression model were mostly similar, with negligible differences in the performance domains tested. The mean ± SDAUCsofthe ﬁnal machine learning models for HDP and GDM were 0 .620 ± 0.001 and 0.671 ± 0.001, respectively.
Conclusions The incidence of HDP and GDM in twin pregnancies in the USA is increasing. The predictive accuracy of the machine learning models for HDP and GDM in

FAQs

What is the study’s main objective?

To develop prediction models for hypertensive disorders of pregnancy (HDP) and gestational diabetes mellitus (GDM) in twin pregnancies using first-prenatal-visit characteristics.

How were the prediction models built and evaluated?

Associations were tested using univariable and multivariable logistic regression. Prediction models used generalized linear models (logit link) and an XGboost machine-learning approach, evaluated with repeated 2-fold cross-validation and AUC.

What did the results show about model performance?

Machine learning and logistic regression had mostly similar predictive performance with negligible differences across tested domains. Models showed modest accuracy and were well calibrated without poor fit.

Aug 5, 2026

Upload by Aurora