Prediction Model of Adverse Pregnancy Outcome in Pre-Eclampsia Based on Logistic Regression and Random Forest...

Area of Science:

• Obstetrics and Gynecology
• Medical Informatics
• Clinical Prediction Modeling

Background:

• Preeclampsia (PE) poses significant risks for adverse pregnancy outcomes.
• Accurate identification of high-risk PE patients is crucial for effective management and intervention.
• Existing prediction models may require enhancement for improved accuracy.

Purpose of the Study:

• To develop and evaluate a prediction model for adverse pregnancy outcomes in preeclampsia (PE).
• To assist clinicians in identifying PE patients at high risk of adverse outcomes.
• To provide guidance for timely treatment interventions.

Main Methods:

• A retrospective study of 319 PE patients was conducted.
• Patients were categorized into adverse (93) and non-adverse (226) outcome groups.
• Logistic regression and random forest models were constructed and compared using a 7:3 training/testing split.

Main Results:

• Key predictors for adverse outcomes included age, small gestational age, clinical symptoms, 24-hour proteinuria, platelet count (PLT), AST, and D-Dimer levels.
• The random forest model demonstrated superior predictive performance compared to the logistic regression model in the test set.
• Significant influencing factors for adverse pregnancy outcomes in PE patients were identified.

Conclusions:

• The random forest model exhibits robust stability and superior prediction efficiency for adverse pregnancy outcomes in PE.
• This model can enhance clinical decision-making for managing high-risk preeclampsia pregnancies.
• Further validation of the random forest model in diverse clinical settings is warranted.