Machine learning-based prediction model for risk factors associated with conversion to open surgery in pediatric laparoscopic radical resection of choledochal cysts.
AI interpretation is pending for this paper.
Open original publication →What the AI sees
Not AI summarized yet.
Research significance
Pending deeper interpretation.
Source abstract
OBJECTIVE: To identify risk factors for conversion to open surgery during laparoscopic choledochal cyst excision in pediatric patients and develop a predictive model using machine learning. METHODS: We retrospectively analyzed 214 children who underwent laparoscopic excision between 2015 and 2023. Patients were classified into laparoscopic (n = 173) or conversion (n = 41) groups. Seven machine learning models were developed and evaluated using AUC and decision curve analysis (DCA). SHAP analysis identified top predictors, which were validated via multivariable logistic regression. A nomogram was constructed and internally validated for discrimination, calibration, and clinical utility. RESULTS: GBM performed best (AUC = 0.897) with the highest net benefit on DCA. SHAP highlighted age, cyst wall thickening (≥ 3 mm), obstructive jaundice, cyst diameter, ALT, and history of perforation. Multivariable analysis confirmed age (OR = 1.022, 95% CI 1.01–1.04; P = 0.004), cyst wall thickening (OR = 15.024, 95% CI 5.03–49.48; P < 0.001), and obstructive jaundice (OR = 10.933, 95% CI 2.76–49.28; P = 0.001) as independent predictors. The nomogram achieved a C-index of 0.843 (95% CI 0.814–0.872), showed good calibration (P = 0.318), and demonstrated consistent performance in cross-validation and DCA. The optimal age cutoff was 27.5 months. CONCLUSION: Age > 27.5 months, cyst wall thickening, and obstructive jaundice are key predictors of conversion. The nomogram provides a clinically useful tool for preoperative risk stratification, pending external validation.