Related Experiment Video
Updated: Oct 30, 2025
04:14
Laparoscopic Choledochal Cyst Excision and Roux-en-Y Choledochojejunostomy in Adults
Published on: February 28, 2025
334
Creation of a Pediatric Choledocholithiasis Prediction Model
Reuven Zev Cohen1, Hongzhen Tian2, Cary G Sauer1
1Division of Pediatric Gastroenterology, Hepatology, and Nutrition, Emory University School of Medicine, Children's Healthcare of Atlanta.
Journal of Pediatric Gastroenterology and Nutrition
|July 5, 2021
Summary
A new model predicts pediatric choledocholithiasis using lab tests and ultrasound. This highly specific tool helps identify children likely to benefit from ERCP, reducing unnecessary procedures.
Area of Science:
- Pediatric Gastroenterology
- Medical Diagnostics
- Machine Learning in Healthcare
Background:
- Pediatric choledocholithiasis diagnosis can be challenging.
- Non-invasive detection is crucial for efficient patient selection for therapeutic ERCP.
- Current methods may lead to unnecessary procedures.
Purpose of the Study:
- Develop a predictive model for pediatric choledocholithiasis.
- Utilize common serum laboratory values and ultrasound results.
- Improve identification of patients benefiting from ERCP.
Main Methods:
- Retrospective review of 316 pediatric patients with suspected choledocholithiasis.
- Multivariate logistic regression and supervised machine learning for model creation.
- Monte-Carlo cross-validation for model validation.
- Determined a specificity threshold of 90% to minimize non-therapeutic ERCP.
Main Results:
- Key predictors identified: ALT, total bilirubin, alkaline phosphatase, and common bile duct diameter.
- The model achieved 90.3% specificity, 40.8% sensitivity, and 71.5% accuracy.
- Positive predictive value was 71.4%, and negative predictive value was 72.1%.
Conclusions:
- A novel predictive model for pediatric choledocholithiasis was developed.
- The model is a highly specific tool for suggesting ERCP.
- Aims to optimize ERCP utilization in pediatric patients.
