Frontiers in pediatrics · 2026
Machine learning early risk assessment model for acute kidney injury in critically ill children: a retrospective cohort study.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
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.
Xie Linyao, Chen Chao, Zhang Chaojie, Chen Lizhi, Li Yijuan
Аннотация
BACKGROUND: Acute kidney injury (AKI) is a common severe complication in intensive care unit (ICU). However, an early risk assessment model that can accurately and promptly predict the risk of AKI in critically ill children remains lacking.
METHODS: This retrospective study included 3,799 children from the Pediatric Intensive Care (PIC) database. The dataset was randomly divided into training set and validation set at a ratio of 7:3. LASSO regression and the Boruta algorithm were employed for feature selection, and the selected variables were incorporated into five machine learning models (Logistic Regression, Random Forest, XGBoost, LightGBM, Support Vector Machine) for training and construction. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), and the SHAP framework was applied for interpretability analysis of the optimal model.
RESULTS: On the validation set, the XGBoost model demonstrated the best risk stratification performance among all five algorithms. SHAP analysis identified bicarbonate, magnesium, activated partial thromboplastin time, lymphocyte count, and thrombin time as the five most important features contributing to the model's predictions.
CONCLUSION: We successfully developed an AKI risk stratification model based on early available clinical data. The model demonstrated acceptable discriminative ability and clinical interpretability in critically ill children, offering potential support for early intervention and improving prognosis.
METHODS: This retrospective study included 3,799 children from the Pediatric Intensive Care (PIC) database. The dataset was randomly divided into training set and validation set at a ratio of 7:3. LASSO regression and the Boruta algorithm were employed for feature selection, and the selected variables were incorporated into five machine learning models (Logistic Regression, Random Forest, XGBoost, LightGBM, Support Vector Machine) for training and construction. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), and the SHAP framework was applied for interpretability analysis of the optimal model.
RESULTS: On the validation set, the XGBoost model demonstrated the best risk stratification performance among all five algorithms. SHAP analysis identified bicarbonate, magnesium, activated partial thromboplastin time, lymphocyte count, and thrombin time as the five most important features contributing to the model's predictions.
CONCLUSION: We successfully developed an AKI risk stratification model based on early available clinical data. The model demonstrated acceptable discriminative ability and clinical interpretability in critically ill children, offering potential support for early intervention and improving prognosis.