Development of a Machine Learning Model to Predict Epidural-Related Maternal Fever During Labor Analgesia: A...

Area of Science:

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

Background:

• Epidural analgesia is common during labor.
• Epidural-related maternal fever (ERMF) is a frequent complication.
• Accurate prediction of ERMF risk is needed for improved patient management.

Purpose of the Study:

• To develop and validate a machine learning (ML) model for predicting ERMF risk.
• To identify key predictors of ERMF.
• To create a tool for real-time clinical risk assessment.

Main Methods:

• Prospective cohort study of 422 parturients receiving epidural labor analgesia.
• Eleven ML algorithms were trained and validated.
• Performance assessed using AUC, AUPRC, accuracy, precision, recall, and F1-score.
• SHapley Additive exPlanations (SHAP) used for feature importance.

Main Results:

• ERMF incidence was 28.1%.
• The CatBoost algorithm achieved the highest performance (AUC=0.94).
• Key predictors included prolonged rupture of membranes, higher BMI, and nulliparity.
• An interactive web tool was developed.

Conclusions:

• A highly discriminative ML model for ERMF risk prediction was developed.
• The CatBoost model effectively identifies high-risk parturients.
• The tool supports evidence-based management of intrapartum fever.