This study aims to develop and internally validate machine learning models to predict surgical timing and discharge duration in patients with infantile hypertrophic pyloric stenosis (IHPS), using admission clinical, biochemical, and ultrasonographic data.

A retrospective analysis was conducted on 55 IHPS cases who underwent pyloromyotomybetween 2015 and 2025. Demographic characteristics, biochemical parameters (blood gas, electrolytes, bilirubin, and urinalysis), and ultrasonographic measurements (pyloric wall thickness, transverse diameter, length, pyloric index [wall thickness/length], and pyloric volume [π/6 × length × diameter²]) were evaluated. Three datasets were constructed to assess different variable combinations: (1) demographic and imaging variables, (2) demographic and laboratory variables, and (3) a combined dataset including all variables. The target outcomes were: (a) surgery performed within 2 days of admission, and (b) discharge within 3 days postoperatively. Class imbalance was addressed using SMOTE and class-weighting strategies. For each algorithm (support vector machine [SVM] and random forest [RF]), four model types were created: standard, SMOTE-enhanced, class weight-adjusted, and ensemble (including extra trees). Models were trained using 4-fold cross-validation repeated 10 times. Performance metrics included F1 score, accuracy, sensitivity, and ROC AUC. Feature selection was performed using the LASSO method to identify the most predictive variables.

ROC AUC values ranged between 0.59 and 0.73 across datasets, indicating moderate discrimination. Pyloric wall thickness and arterial blood gas pH emerged as the strongest predictors of early surgery, whereas pyloric transverse diameter and urine specific gravity were most strongly associated with early discharge.

Machine learning models demonstrated moderate and exploratory predictive performance for early perioperative outcomes in IHPS. These findings should be interpreted as hypothesis-generating rather than decision-guiding and require external validation before clinical implementation.