EUVIMEDEUVIMED entdecken
KohortenstudieOpen Access · CC BY 4.0

First-Trimester Gestational Diabetes Mellitus Risk Prediction with Machine Learning Techniques: Results from the BORN2020 Cohort Study

Vorhersage des Gestationsdiabetes im ersten Trimenon mit Machine-Learning-Techniken: Ergebnisse der BORN2020-Kohortenstudie

Nikolaos Pazaras iD, Antonios Siargkas iD, Antigoni Tranidou iD, Aikaterini Apostolopoulou iD, Ioannis Tsakiridis iD, Panagiotis D Bamidis iD, Sofoklis Stavros, Anastasios Potiris iD, Michail Chourdakis iD, Themistoklis Dagklis iD

Journal of Clinical Medicine · 23.03.2026 · 15 (6) · 2461

DOI: 10.3390/jcm15062461 ↗

Auf einen Blick

Eine prospektive Kohortenstudie mit 797 Schwangeren prüfte ML-Modelle zur Vorhersage von GDM vor dem oralen Glukosetoleranztest. Die beste Modellleistung war nur mäßig, und Routinevariablen schnitten ähnlich ab wie Modelle mit Ernährungsdaten.

Fragestellung

Können ML-Modelle mit klinischen und diätetischen Daten aus dem ersten Trimenon das GDM-Risiko vor dem Standardtest vorhersagen?

Methode

Analyse von 797 Schwangeren der BORN2020-Kohortenstudie; 10 ML-Algorithmen, 5 Strategien zum Umgang mit Klassenungleichgewicht, stratifizierte 5-fold-Cross-Validation und unabhängiger 20 %-Testdatensatz.

Ergebnisse

Bestes Modell: logistische Regression ohne Resampling, AUC-ROC 0,664 (95 %-KI 0,542–0,777), Sensitivität 0,783, NPV 0,932. Ein Modell mit 9 Merkmalen erreichte AUC 0,712 (95 %-KI 0,589–0,825). Kein Modell lag über AUC 0,80.

Einschränkungen

Keine externe Validierung; geringe events-per-variable-Relation (~3,8); deutliche Fehlkalibrierung (Steigung 0,56; Interzept −1,83).

Schlussfolgerung der Autor:innen

Die Autor:innen sehen eine nur mäßige Diskriminierung für die frühe GDM-Vorhersage; Routinevariablen waren vergleichbar mit Modellen inklusive detaillierter Ernährungsdaten. Die Modelle bleiben explorativ und brauchen externe Validierung und Rekalibrierung.

Kernaussagen

  • Beste AUC-ROC: 0,664 (95 %-KI 0,542–0,777)
  • 9-Merkmals-Modell: AUC 0,712 (95 %-KI 0,589–0,825)
  • Keine Verbesserung durch zusätzliche FFQ-Mikronährstoffe

Deutsche Zusammenfassung: KI-gestützt aus dem Original-Abstract erstellt; alle Zahlen wurden automatisch mit dem Original abgeglichen. Maßgeblich ist der Original-Abstract.

Kurz gefasst

Kohortenstudie in der Zeitschrift Journal of Clinical Medicine, online erschienen am 23.03.2026. Verfasst von Nikolaos Pazaras und 9 weiteren Autor:innen. Beteiligte Einrichtung: Third Department of Obstetrics and Gynecology, School of Medicine, Faculty of Health Sciences, Aristotle University of Thessaloniki, 541 24 Thessalon…. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 48 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

Auf einen BlickAbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

Gestational diabetes mellitus (GDM) affects many pregnancies worldwide and is associated with adverse maternal and fetal outcomes. Current screening at 24–28 weeks limits opportunities for early intervention. We evaluated whether machine learning (ML) models using first-trimester clinical and dietary data can predict GDM risk before the standard oral glucose tolerance test.

Methods

We analyzed data from 797 pregnant women enrolled in the BORN2020 prospective cohort study (Thessaloniki, Greece). Ten ML algorithms were evaluated across five class-imbalance handling strategies using stratified 5-fold cross-validation, with final evaluation on an independent 20% held-out test set. Features included maternal demographics, obstetric history, lifestyle factors, and 22 dietary micronutrient intakes from the pre-pregnancy period assessed by Food Frequency Questionnaire.

Results

The best-performing model (Logistic Regression without resampling) achieved an AUC-ROC of 0.664 (95% CI: 0.542–0.777), with sensitivity of 0.783 and NPV of 0.932 at the pre-specified threshold. The high NPV should be interpreted in the context of the low GDM prevalence (14.7%), as NPV is mathematically dependent on disease prevalence. A reduced nine-feature model using only routine clinical and demographic variables achieved a numerically higher AUC of 0.712 (95% CI: 0.589–0.825), with overlapping confidence intervals, indicating that detailed FFQ-derived micronutrient data did not improve prediction. Maternal age and pre-pregnancy BMI were the strongest individual predictors by SHAP analysis. No model reached the AUC >0.80 threshold for good discrimination. Substantial miscalibration was observed (slope: 0.56; intercept: −1.83), limiting use for absolute risk estimation.

Conclusions

This exploratory study demonstrates that first-trimester ML models achieve modest discriminative ability for early GDM prediction, with routine clinical variables performing comparably to models incorporating detailed dietary assessment. These findings should be interpreted with caution, as no external validation cohort was available and the low events-per-variable ratio (~3.8) constrains the reliability of individual model estimates. Substantial miscalibration further limits use for absolute risk estimation. Accordingly, these models should be regarded as exploratory risk-ranking tools only and require external validation and recalibration before any clinical implementation.

Schlagwörter

Stichwörter der Autor:innen

GDMMLSHAPclass imbalanceexplainable AIgestational diabetes mellitusgradient boosting methodsmachine learningmaternal nutritionmicronutrientsprediction

Autor:innen und Institutionen

  • Third Department of Obstetrics and Gynecology, School of Medicine, Faculty of Health Sciences, Aristotle University of Thessaloniki, 541 24 Thessaloniki, Greece.
  • Laboratory of Hygiene, Social & Preventive Medicine and Medical Statistics, School of Medicine, Faculty of Health Sciences, Aristotle University of Thessaloniki, 541 24 Thessaloniki, Greece.
  • Laboratory of Medical Physics and Digital Innovation, School of Medicine, Faculty of Health Sciences, Aristotle University of Thessaloniki (AUTH), 541 24 Thessaloniki, Greece.
  • Third Department of Obstetrics and Gynecology, University General Hospital "ATTIKON", Medical School, National and Kapodistrian University of Athens, 124 62 Athens, Greece.

ORCID-Profile: Nikolaos Pazaras, Antonios Siargkas, Antigoni Tranidou, Aikaterini Apostolopoulou, Ioannis Tsakiridis, Panagiotis D Bamidis, Anastasios Potiris, Michail Chourdakis, Themistoklis Dagklis

Zitieren

APA

Pazaras, N., Siargkas, A., Tranidou, A., Apostolopoulou, A., Tsakiridis, I., Bamidis, P. D., Stavros, S., Potiris, A., Chourdakis, M., & Dagklis, T. (2026). First-Trimester Gestational Diabetes Mellitus Risk Prediction with Machine Learning Techniques: Results from the BORN2020 Cohort Study. Journal of Clinical Medicine, 15(6), 2461. https://doi.org/10.3390/jcm15062461

Vancouver

Pazaras N, Siargkas A, Tranidou A, Apostolopoulou A, Tsakiridis I, Bamidis PD, et al. First-Trimester Gestational Diabetes Mellitus Risk Prediction with Machine Learning Techniques: Results from the BORN2020 Cohort Study. Journal of Clinical Medicine. 2026;15(6):2461. doi:10.3390/jcm15062461

BibTeX
@article{Pazaras2026,
  author = {Pazaras, Nikolaos and Siargkas, Antonios and Tranidou, Antigoni and Apostolopoulou, Aikaterini and Tsakiridis, Ioannis and Bamidis, Panagiotis D and Stavros, Sofoklis and Potiris, Anastasios and Chourdakis, Michail and Dagklis, Themistoklis},
  title = {First-Trimester Gestational Diabetes Mellitus Risk Prediction with Machine Learning Techniques: Results from the BORN2020 Cohort Study},
  journal = {Journal of Clinical Medicine},
  year = {2026},
  volume = {15},
  number = {6},
  pages = {2461},
  doi = {10.3390/jcm15062461}
}
Kostenlos starten →