Gestational diabetes & diabetes in pregnancy
Predicting postpartum glucose intolerance in women with gestational diabetes mellitus in primary care: A machine learning approach using XGBoost and SHAP values
An erratum to this article was published (Diabetes Res Clin Pract, Mar 2026 ↗)
Authors' conclusion
An interpretable XGBoost model with SHAP explanations using routine antenatal data shows promise for postpartum glucose risk assessment in primary care. Despite moderate predictive performance, the model demonstrated a high negative predictive value.
Study context
- Population
- Women with GDM who completed a 6-week postpartum 75-g OGTT, primary care
- Participants
- 600 women
- Outcome
- Prediction of postpartum glucose intolerance (model performance: AUC, precision, recall, F1-score, NPV)
- Follow-up
- 6 weeks postpartum
Methodological detailsmachine learning (XGBoost) with SHAP, stratified 5-fold cross-validation, ROSE oversampling, grid search optimisation, primary care
- Machine learning (XGBoost) with SHAP
- Stratified 5-fold cross-validation
- ROSE oversampling
- Grid search optimisation
- Primary care
Performance
As reported in the abstract
Model performance for postpartum glucose intolerance
| Test or model | AUC | PR-AUC | Precision | Recall | NPV |
|---|---|---|---|---|---|
| XGBoost | 0.671 | 0.35 | 0.79 | 0.82 | 0.87 |
XGBoost
AUC0.671
PR-AUC0.35
Precision0.79
Recall0.82
NPV0.87
Further findings
Incidence of postpartum glucose intolerance · 6 weeks postpartum
- Postpartum glucose intolerance
- 19%
Key predictors (SHAP)
- Key predictors: FPG, 2-hour glucose, gestational weight gain, multiparity, previous GDM, family history of diabetes
Original publication
Predicting postpartum glucose intolerance in women with gestational diabetes mellitus in primary care: A machine learning approach using XGBoost and SHAP values. Diabetes Res Clin Pract. Published 12 January 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →