Frontiers in cellular and infection microbiology · 2026
Machine learning prediction of sepsis in paralytic ileus using interpretable clinical models.
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.
Song Qingzhou, Wu Xuanlin, Taheri Firooz Ahmad, Meng Linghou, Wang Wentao, Mo Xianwei
Аннотация
BACKGROUND: Paralytic ileus (PI) is a common complication in critically ill patients, and the development of sepsis significantly worsens prognosis. Early identification of sepsis risk in PI patients remains a clinical challenge.
METHODS: Adult PI patients were identified from the MIMIC-IV database (2008-2022) and randomly split into a training cohort (70%) and an internal validation cohort (30%). An independent external cohort was obtained from Guangxi Medical University Cancer Hospital. Predictors were extracted within the first 12 hours after admission. Feature selection was performed using LASSO regression and the Boruta algorithm combined with clinical knowledge. Seven machine learning models were evaluated. Model performance was assessed by area under the receiver operating characteristic curve (AUC), calibration, and decision curve analysis (DCA). SHAP was applied for interpretability.
RESULTS: A total of 579 PI patients were included, of whom 29.78% developed new-onset sepsis. Seven predictors were retained: pneumonia, red cell distribution width (RDW), heart failure, blood urea nitrogen (BUN), atrial fibrillation, serum chloride, and white blood cell count. Logistic regression demonstrated stable performance and was selected as the final model, achieving an AUC of 0.687 in internal validation and 0.715 in external validation. Calibration and DCA indicated good agreement and consistent clinical net benefit. SHAP analysis identified pneumonia and RDW as the most influential predictors.
CONCLUSION: An early, interpretable logistic regression model based on readily available clinical variables can effectively predict new-onset sepsis in PI patients and may support timely risk stratification and preventive intervention.
METHODS: Adult PI patients were identified from the MIMIC-IV database (2008-2022) and randomly split into a training cohort (70%) and an internal validation cohort (30%). An independent external cohort was obtained from Guangxi Medical University Cancer Hospital. Predictors were extracted within the first 12 hours after admission. Feature selection was performed using LASSO regression and the Boruta algorithm combined with clinical knowledge. Seven machine learning models were evaluated. Model performance was assessed by area under the receiver operating characteristic curve (AUC), calibration, and decision curve analysis (DCA). SHAP was applied for interpretability.
RESULTS: A total of 579 PI patients were included, of whom 29.78% developed new-onset sepsis. Seven predictors were retained: pneumonia, red cell distribution width (RDW), heart failure, blood urea nitrogen (BUN), atrial fibrillation, serum chloride, and white blood cell count. Logistic regression demonstrated stable performance and was selected as the final model, achieving an AUC of 0.687 in internal validation and 0.715 in external validation. Calibration and DCA indicated good agreement and consistent clinical net benefit. SHAP analysis identified pneumonia and RDW as the most influential predictors.
CONCLUSION: An early, interpretable logistic regression model based on readily available clinical variables can effectively predict new-onset sepsis in PI patients and may support timely risk stratification and preventive intervention.