Development of a prediction model on preeclampsia using machine learning-based method: a retrospective cohort study...

Area of Science:

• Obstetrics and Gynecology
• Medical Informatics
• Computational Biology

Background:

• Preeclampsia (PE) is a significant complication of pregnancy.
• Early detection of PE is crucial for improved maternal and fetal outcomes.
• Predictive models can aid in identifying pregnancies at high risk for PE.

Purpose of the Study:

• To develop predictive models for preeclampsia using machine learning.
• To analyze clinical and laboratory data from early pregnancy screening.
• To identify key predictive features for PE risk.

Main Methods:

• Retrospective review of medical records.
• Application of five machine learning algorithms: deep neural network (DNN), logistic regression (LR), support vector machine (SVM), decision tree (DT), and random forest (RF).
• Incorporation of 18 variables including maternal characteristics, medical history, laboratory, and ultrasound results.

Main Results:

• The random forest (RF) model exhibited the highest predictive accuracy.
• RF model achieved an Area Under the Receiver Operating Curve (AUROC) of 0.86.
• Key predictive features for PE risk were automatically identified by the machine learning approach.

Conclusions:

• Machine learning methods can effectively predict preeclampsia risk from early pregnancy data.
• The developed models demonstrate high predictive performance.
• This approach aids in early identification of pregnancies at risk for PE.