Machine learning-based prediction model for chronic post-surgical pelvic pain syndrome: a comprehensive analysis...

Area of Science:

• Pain Medicine
• Machine Learning in Healthcare
• Surgical Outcomes Research

Background:

• Chronic post-surgical pelvic pain syndrome (CPSPP) is a significant challenge after pelvic surgeries.
• Its complex nature requires advanced predictive methods for better patient outcomes.
• Optimizing treatment strategies for CPSPP is crucial for patient recovery.

Purpose of the Study:

• To develop and validate a machine learning (ML) model for predicting CPSPP treatment outcomes.
• To enhance model interpretability using SHAP (SHapley Additive exPlanations) analysis.
• To identify key risk factors influencing CPSPP treatment response for improved clinical decision-making.

Main Methods:

• Retrospective cohort study of 62 patients undergoing pelvic surgery.
• Collected comprehensive clinical data: demographics, surgical details, pain assessments, outcomes.
• Employed multiple ML algorithms, evaluated with ROC analysis; utilized SHAP for feature importance.

Main Results:

• XGBoost model showed high performance (AUC 0.94 training, 0.77 validation).
• SHAP analysis revealed baseline VAS score, stimulation frequency, and intensity as key predictors.
• Identified 27.4% significant clinical improvement post-magnetic stimulation therapy.

Conclusions:

• A novel, interpretable ML approach for predicting CPSPP treatment outcomes was developed.
• Findings enhance understanding of CPSPP risk factors.
• Provides a foundation for personalized treatment and clinical decision support systems.