Development and Validation of a LASSO-Based Prediction Model for Postoperative Bleeding Following Pediatric Tonsillectomy and Adenoidectomy

AIM: Despite technological advancements, children undergoing adenotonsillectomy using low-temperature plasma radiofrequency ablation remain at risk for postoperative bleeding. This study aimed to construct a precise machine learning-based model to predict the risk of postoperative bleeding in this population. METHODS: A retrospective cohort study involving 306 pediatric patients who underwent the procedure between January 2023 and January 2025 was conducted. Clinical data, including demographic characteristics, preoperative laboratory findings, and surgical details, were collected and cleaned. The dataset was randomly divided into a training set (n = 244) and a validation set (n = 62) at an 8:2 ratio. A predictive model was developed using the least absolute shrinkage and selection operator (LASSO) cross-validation algorithm, and its performance was assessed in the validation cohort. RESULTS: Among 306 children, 24 (7.8%) experienced postoperative bleeding. Univariate comparisons identified significant differences in 11 variables: age, height, weight, body mass index (BMI), disease duration, postoperative infection, dietary compliance, postoperative pain, lymphocyte count, neutrophil-to-lymphocyte ratio (NLR), and neutrophil-to-platelet ratio (NPR) (all p < 0.05). At lambda.1se = 0.022, LASSO logistic regression retained four predictors: postoperative infection, dietary compliance, age, and postoperative pain. After 1000 bootstrap resamples, the out-of-bag area under the curve (AUC) was 0.935 (95% CI: 0.875–0.995), with a Brier score of 0.035, calibration intercept of 0.045, and calibration slope of 1.418. At the optimal probability cutoff of 0.114, sensitivity was 0.875, specificity was 0.936, and the Youden index was 0.811. CONCLUSIONS: Postoperative infection, dietary adherence, postoperative pain, and age are significant predictors of bleeding following pediatric adenotonsillectomy using low-temperature plasma radiofrequency ablation. The LASSO-based model provides a reliable tool for identifying high-risk patients and optimizing perioperative management.