A comprehensive first-trimester predictive model for preeclampsia based on multi-indicators and machine learning: A...

Area of Science:

• Obstetrics and Gynecology
• Maternal-Fetal Medicine
• Biomedical Data Science

Background:

• Preeclampsia (PE) is a severe pregnancy disorder causing significant maternal and perinatal complications.
• Early prediction of PE is crucial for timely intervention and improved outcomes.
• Current prediction methods lack comprehensive early-stage assessment.

Purpose of the Study:

• To develop and validate a predictive model for preeclampsia using first-trimester maternal indicators.
• To identify key biomarkers and employ machine learning for robust PE prediction.
• To assess the model's generalizability in an independent external validation cohort.

Main Methods:

• Retrospective study of 100 pregnant individuals (50 PE, 50 controls) with first-trimester data.
• Feature selection using Least Absolute Shrinkage and Selection Operator (LASSO) regression.
• Development and evaluation of 7 machine learning algorithms, including a neural network model, with external validation.

Main Results:

• LASSO identified 12 key predictive features, including placental growth factor (PlGF), uterine artery pulsatility index (UtAPI), C-reactive protein (CRP), and neutrophil-to-lymphocyte ratio (NLR).
• The neural network model achieved an area under the curve (AUC) of 0.917 in internal validation and 0.838 in external validation.
• SHapley Additive exPlanations confirmed PlGF, UtAPI, CRP, and NLR as highly influential predictors.

Conclusions:

• A robust predictive model for preeclampsia was developed using early pregnancy biomarkers and machine learning.
• The neural network model demonstrates superior discriminative ability for PE prediction.
• Early identification via this model can facilitate timely interventions, potentially improving maternal and fetal health.