Prediction Models for Late-Onset Preeclampsia: A Study Based on Logistic Regression, Support Vector Machine, and...

Area of Science:

• Obstetrics and Gynecology
• Medical Informatics
• Machine Learning in Healthcare

Background:

• Preeclampsia affects 2-4% of pregnancies globally, with late-onset preeclampsia being common.
• Existing prediction models lack early detection capabilities and often use inaccessible indicators.
• This limits applicability in resource-limited settings.

Purpose of the Study:

• To develop and evaluate prediction models for late-onset preeclampsia.
• Utilize general information, maternal risk factors, and early gestation laboratory indicators (6-13 weeks).

Main Methods:

• Analysis of 2000 pregnancies (110 with late-onset preeclampsia).
• Data included hospital information system data and early pregnancy laboratory results.
• Compared logistic regression, support vector machine (SVM), and extreme gradient boosting (XGBoost) models.

Main Results:

• SVM and XGBoost models significantly improved late-onset preeclampsia detection rates compared to logistic regression.
• SVM showed a higher false positive rate; logistic regression and XGBoost had high negative predictive values (99.3%).
• Logistic regression achieved the highest area under the ROC curve (0.877), indicating strong predictive advantages.

Conclusions:

• SVM and XGBoost offer improved detection for late-onset preeclampsia using early pregnancy data.
• Logistic regression remains a strong predictive model, especially for identifying women at low risk.
• The study highlights the potential of machine learning and traditional models for early preeclampsia prediction.