Related Experiment Videos
Development of an Interpretable Machine Learning Model for Predicting Clavien-Dindo Grade ≥2 Complications after
Haotian Pan1,2,3,4, Runwu Wang1,2,3,4, Mengnan Jiang1,2,3,4
1Department of Urology, Traditional Chinese Medicine Innovation Team, Laboratory for Targeted Delivery Application of Traditional Chinese Medicine, Ministry of Education Key Laboratory of Child Development and Disorders, Chongqing, P.R. China.
Journal of Endourology
|June 23, 2026
Summary
Machine learning accurately predicts complications after minimally invasive pyeloplasty in children. This tool helps surgeons assess risks for pediatric ureteropelvic junction obstruction (UPJO) patients, enabling personalized follow-up strategies.
Area of Science:
- Pediatric Urology
- Surgical Outcomes
- Machine Learning in Medicine
Background:
- Minimally invasive pyeloplasty (MIP) is standard for pediatric ureteropelvic junction obstruction (UPJO).
- Postoperative complications significantly impact surgical outcomes in MIP procedures.
- Predicting these complications preoperatively is crucial for patient management.
Purpose of the Study:
- To develop a machine learning model for predicting Clavien-Dindo grade ≥2 complications after pediatric MIP.
- To identify key preoperative predictors of postoperative complications.
- To create a user-friendly tool for real-time clinical risk stratification.
Main Methods:
- Retrospective analysis of 533 pediatric UPJO patients.
- Recursive feature elimination for feature selection.
- Development and comparison of ten machine learning algorithms, including LightGBM.
Main Results:
- 127 (23.8%) patients experienced Clavien-Dindo grade ≥2 complications.
- Key predictors identified: age, kidney volume ratio, calyx diameters, neutrophil count, cystatin C, pelvic diameter, and WBC count.
- LightGBM model showed superior predictive performance, validated by SHAP analysis and an online calculator.
Conclusions:
- The LightGBM model effectively predicts complications following unilateral MIP.
- This tool aids clinicians in risk assessment and personalized follow-up planning.
- Facilitates early intervention for improved pediatric UPJO management.
