Machine learning insights on optimal timing for re-suturing postpartum perineal wound dehiscence

Area of Science:

• Obstetrics and Gynecology
• Surgical Wound Management
• Medical Informatics

Background:

• Postpartum perineal wound dehiscence presents a significant clinical challenge.
• Standardized decision-making for re-suturing is often lacking.

Purpose of the Study:

• To identify key factors influencing re-suturing decisions for perineal wound dehiscence.
• To develop and validate machine learning models for predicting re-suturing needs.

Main Methods:

• Analysis of 129 cases of postpartum perineal wound dehiscence using five machine learning models: Logistic Regression, Decision Tree, Random Forest, Support Vector Machine (SVM), and Extreme Gradient Boosting (XGBoost).
• Evaluation of variables including clinical symptoms, wound characteristics, perineal trauma severity, and clinician seniority.
• Model validation using a 20% hold-out data set.

Main Results:

• The Support Vector Machine (SVM) model demonstrated superior performance (AUC 0.86, accuracy 0.86), identifying wound depth, length, and pain score as primary predictors.
• Decision Tree and XGBoost models achieved 0.82 accuracy, highlighting subspecialist involvement, wound depth, and discharge as important features.
• Wound depth and clinician seniority consistently emerged as critical predictors across all models.

Conclusions:

• Machine learning models reliably predict the need for re-suturing in perineal wound dehiscence.
• Wound severity and clinician seniority are the strongest predictors, supporting data-driven standardization of care.
• Future research should focus on prospective validation and clinical integration of these predictive models.