What does this research mean for the field?

Machine learning models, particularly XGBoost, effectively predict emergency delivery within 48 hours of early-onset preeclampsia diagnosis with high discrimination (AUROC 0.908), with CRP, D-dimer, and hypoproteinemia serving as key predictors. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

November 6, 2025BMC Medical Informatics and Decision MakingOpen Access

Risk stratification and prediction of emergency delivery in early-onset preeclampsia using machine learning

Q: What is the clinical evidence from this study?

Study design: Cohort. Population: Early-onset preeclampsia (n=648). Intervention: XGBoost machine learning model vs. Other machine learning models. Primary outcome: Prediction of emergency delivery (≤ 48 h post-diagnosis) (AUROC 0.908).

View Full Paper

Ask AI

Bookmark

Key result

XGBoost model predicts 48-hour emergency delivery in early-onset preeclampsia with ~0.91 AUROC.

AUROC 0.908
n=648

Why the study?

Early-onset preeclampsia poses significant risks for maternal and fetal outcomes, particularly when emergency delivery is required, necessitating improved prediction methods.

Population

648 singleton pregnancies diagnosed with early-onset preeclampsia at 28-34 weeks gestation

Comparison

Emergency delivery within 48 hours post-diagnosis vs non-emergency delivery

Design

Retrospective cohort study with machine learning model development and evaluation

Follow-up

48 hours post-diagnosis

Authors

YXYanhong Xu XLXinying Liu YZYing Zhang

Discussion

Member takes

Overview

May support risk stratification for emergency delivery in early-onset preeclampsia; leaves open prospective validation before clinical use.

Study Design

Type

Cohort (n=648)

Multicenter

Structured PICO

Population

648 pregnant women with early-onset preeclampsia and singleton pregnancies at 28 to 34 weeks gestation, retrospectively analyzed to predict the need for emergency delivery within 48 hours of diagnosis.

Exposure

Machine learning predictive models (including XGBoost, GBDT, logistic regression, naïve Bayes, LightGBM, SVM, MLP, elastic net) using 16 clinical and biochemical predictors

Outcome

Emergency delivery (≤ 48 h post-diagnosis)

Main Result

Effect estimate: AUROC 0.908

Machine learning models, particularly XGBoost, can accurately predict the need for emergency delivery within 48 hours in early-onset preeclampsia using routine clinical and laboratory features.

Limitations

Retrospective design might introduce selection bias, necessitating prospective validation.
Single-center sample requires external validation with multi-center data.
Emerging biomarkers (e.g., placental growth factor) were not included.
Retrospective design
Single-center study
Lack of prospective validation
PlGF (Placental Growth Factor) not included as it is not part of standard workflow

Cite This Study

Xu et al. (2025) conducted a cohort in Early-onset preeclampsia (n=648). XGBoost machine learning model vs. Other machine learning models was evaluated on Prediction of emergency delivery (≤ 48 h post-diagnosis) (AUROC 0.908). An XGBoost machine learning model effectively predicted emergency delivery within 48 hours in early-onset preeclampsia with a testing AUROC of 0.908, identifying CRP, D-dimer, and hypoproteinemia as key predictors.

synapsesocial.com/papers/6aacbb4e47564697cdb0dce8 https://doi.org/10.1186/s12911-025-03249-4

View Full Paper

Ask AI

Bookmark

November 6, 2025BMC Medical Informatics and Decision MakingOpen Access

Risk stratification and prediction of emergency delivery in early-onset preeclampsia using machine learning

View Full Paper

Ask AI

Bookmark

Key result

XGBoost model predicts 48-hour emergency delivery in early-onset preeclampsia with ~0.91 AUROC.

AUROC 0.908
n=648

Why the study?

Early-onset preeclampsia poses significant risks for maternal and fetal outcomes, particularly when emergency delivery is required, necessitating improved prediction methods.

Population

648 singleton pregnancies diagnosed with early-onset preeclampsia at 28-34 weeks gestation

Comparison

Emergency delivery within 48 hours post-diagnosis vs non-emergency delivery

Design

Retrospective cohort study with machine learning model development and evaluation

Follow-up

48 hours post-diagnosis

Authors

YXYanhong Xu XLXinying Liu YZYing Zhang

Discussion

Member takes

Overview

May support risk stratification for emergency delivery in early-onset preeclampsia; leaves open prospective validation before clinical use.

Study Design

Type

Cohort (n=648)

Multicenter

Structured PICO

Population

Exposure

Machine learning predictive models (including XGBoost, GBDT, logistic regression, naïve Bayes, LightGBM, SVM, MLP, elastic net) using 16 clinical and biochemical predictors

Outcome

Emergency delivery (≤ 48 h post-diagnosis)

Main Result

Effect estimate: AUROC 0.908

Machine learning models, particularly XGBoost, can accurately predict the need for emergency delivery within 48 hours in early-onset preeclampsia using routine clinical and laboratory features.

Limitations

Retrospective design might introduce selection bias, necessitating prospective validation.
Single-center sample requires external validation with multi-center data.
Emerging biomarkers (e.g., placental growth factor) were not included.
Retrospective design
Single-center study
Lack of prospective validation
PlGF (Placental Growth Factor) not included as it is not part of standard workflow

Cite This Study

synapsesocial.com/papers/6aacbb4e47564697cdb0dce8 https://doi.org/10.1186/s12911-025-03249-4

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

View Full Paper

Ask AI

Bookmark