Pediatric surgery international · 2026
Machine learning-based prediction model for postoperative blood transfusion in pediatric patients with Hirschsprung's disease.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Luo Jianfeng, Wei Yangyi, Zhou Ruijie, Huang Kaikun, Li Qi, Li Long
Аннотация
BACKGROUND: Children undergoing Hirschsprung's disease (HD) surgery are at risk of postoperative transfusion, which may cause complications and increase resource use. Current decisions are often empirical and lack individualized prediction tools.
METHODS: We retrospectively analyzed 889 pediatric patients who underwent HD corrective surgery at a tertiary center (2015-2025). Postoperative transfusion was defined as ≥ 0.5 units of packed red blood cells within 72 h after surgery. Forty-seven preoperative variables were collected; least absolute shrinkage and selection operator (LASSO) regression identified key predictors. Seven machine learning models, including support vector machine (SVM), were developed and evaluated by AUC, sensitivity, specificity, calibration, and decision curve analysis (DCA). Model interpretability was assessed using SHAP analysis.
RESULTS: Of 889 patients, 155 (17.4%) required postoperative transfusion, while intraoperative transfusions were rare. LASSO selected five predictors: hemoglobin, HD classification, red blood cell count, prothrombin time, and total protein. The SVM model achieved the best performance in testing cohort (AUC 0.8361; accuracy 77.15%; sensitivity 76.60%; specificity 77.27%; negative predictive value 93.92%). Calibration and DCA supported good reliability and clinical net benefit. Hemoglobin was the most influential predictor.
CONCLUSION: We developed an interpretable preoperative machine learning model for predicting transfusion risk after HD surgery in children. The SVM model demonstrated good discrimination and may assist in perioperative planning, particularly for identifying low-risk patients who may not require routine crossmatching. External multi-center validation is needed before clinical implementation.
METHODS: We retrospectively analyzed 889 pediatric patients who underwent HD corrective surgery at a tertiary center (2015-2025). Postoperative transfusion was defined as ≥ 0.5 units of packed red blood cells within 72 h after surgery. Forty-seven preoperative variables were collected; least absolute shrinkage and selection operator (LASSO) regression identified key predictors. Seven machine learning models, including support vector machine (SVM), were developed and evaluated by AUC, sensitivity, specificity, calibration, and decision curve analysis (DCA). Model interpretability was assessed using SHAP analysis.
RESULTS: Of 889 patients, 155 (17.4%) required postoperative transfusion, while intraoperative transfusions were rare. LASSO selected five predictors: hemoglobin, HD classification, red blood cell count, prothrombin time, and total protein. The SVM model achieved the best performance in testing cohort (AUC 0.8361; accuracy 77.15%; sensitivity 76.60%; specificity 77.27%; negative predictive value 93.92%). Calibration and DCA supported good reliability and clinical net benefit. Hemoglobin was the most influential predictor.
CONCLUSION: We developed an interpretable preoperative machine learning model for predicting transfusion risk after HD surgery in children. The SVM model demonstrated good discrimination and may assist in perioperative planning, particularly for identifying low-risk patients who may not require routine crossmatching. External multi-center validation is needed before clinical implementation.