JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

Application of machine learning models to identify predictors of good outcome after laparoscopic fundoplication

  • 0Flinders University Discipline of Surgery, College of Medicine and Public Health, Flinders Medical Centre, Bedford Park, South Australia, Australia.
Journal of Gastrointestinal Surgery : Official Journal of the Society for Surgery of the Alimentary Tract +

|

March 23, 2025

|

March 23, 2025

View abstract on PubMed

Summary

This summary is machine-generated.

Predicting outcomes after laparoscopic fundoplication is challenging. While male sex and esophageal peristalsis are key predictors, machine learning models offer only marginal improvements in accuracy for gastroesophageal reflux disease treatment.

Area Of Science

  • Gastroenterology
  • Surgical Outcomes
  • Medical Informatics

Background

  • Laparoscopic fundoplication is the standard treatment for gastroesophageal reflux disease (GERD).
  • 10-20% of patients experience persistent symptoms post-surgery, necessitating further intervention.
  • This study investigated predictors of outcomes following laparoscopic fundoplication.

Purpose Of The Study

  • To identify preoperative factors influencing postoperative outcomes after laparoscopic fundoplication.
  • To evaluate the predictive accuracy of regression and machine learning (ML) models for GERD symptom resolution and patient satisfaction.
  • To determine the clinical utility of advanced predictive modeling in managing GERD patients.

Main Methods

  • Analysis of prospectively maintained data from 894 patients undergoing primary laparoscopic fundoplication (1998-2015).
  • Utilized regression and machine learning models to assess preoperative factors impacting heartburn, dysphagia, and satisfaction scores.
  • Median follow-up was 5 years, with specific analysis on complete datasets (n=221) and imputation methods.

Main Results

  • Machine learning (least absolute shrinkage support operator) and negative binomial regression showed similar accuracy in predicting heartburn scores (RMSE ≈ 2.3-2.4).
  • Male sex was a significant predictor for reduced heartburn and dysphagia.
  • Percentage of esophageal peristalsis significantly predicted satisfaction and reduced dysphagia.

Conclusions

  • Male sex and esophageal peristalsis are significant predictors of outcomes after laparoscopic fundoplication.
  • Current regression and ML models offer limited accuracy in predicting individual patient outcomes.
  • Clinical judgment and patient counseling remain paramount, as predictive models do not replace physician expertise for managing GERD.

Keywords: