European journal of clinical microbiology & infectious diseases : official publication of the European Society of Clinical Microbiology · 2025
Construction and validation of a predictive model for the risk of multidrug-resistant Klebsiella pneumoniae infection based on machine learning algorithms - a multicenter retrospective study.
СтатьяМногоцентровое исследованиеВалидационное исследованиеНазвание 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.
Sun Tao, Wang Pei-Pei, Liu Jun-Ji, An Zhen, Zhao Jun-Rong, Liu Jun
Аннотация
OBJECTIVE: The development of a reliable predictive model for Multi-drug-resistant Klebsiella pneumoniae (MDR-KP) infections is imperative for the timely identification of at-risk individuals.
METHODS: This study analyzed data from 3,554 patients with KP infection at multiple hospitals. By comparing six machine learning algorithms (Logistic Regression(LR), Efficient Neural Network (ENet), Decision Tree(DT), MultiLayer Perceptron(MLP), Support Vector Machine(SVM), and Extreme Gradient Boosting(XGBoost)), we constructed and validated the prediction model. Furthermore, the model interpretation was conducted through SHapley Additive exPlanations (SHAP) analysis. Subsequently, nomograms were developed to estimate the risk of MDR-KP infection in hospitalized individuals. Finally, the association between independent variables and the risk of MDR-KP acquisition was elucidated through Restricted Cubic Spline (RCS) analysis.
RESULTS: The results of the multivariable logistic regression analysis indicated that C-reactive protein (CRP), Uric Acid (UA), Urea, Platelet (PLT), Hemoglobin (HB), Red blood cell counts (RBC), Age, and Gender were identified as independent risk factors. We incorporated these independent risk factors into six machine learning, and found that the XGBoost-based model exhibited superior performance compared to other machine learning algorithms, achieving a recall of 0.732, an F1 score of 0.707, and an AUC of 0.777. Furthermore, the SHAP method highlighted Urea, UA, and PLT as the primary decision factors predicted by the machine learning model, and the RCS analysis revealed a nonlinear relationship between Age, CRP, RBC, HB, UA, UREA, and the risk of MDR-KP infection.
CONCLUSION: This study has developed an effective risk prediction model for MDR-KP infection. The model has the potential to assist healthcare providers in early identification of high-risk patients, enabling timely implementation of preventive and therapeutic interventions.
METHODS: This study analyzed data from 3,554 patients with KP infection at multiple hospitals. By comparing six machine learning algorithms (Logistic Regression(LR), Efficient Neural Network (ENet), Decision Tree(DT), MultiLayer Perceptron(MLP), Support Vector Machine(SVM), and Extreme Gradient Boosting(XGBoost)), we constructed and validated the prediction model. Furthermore, the model interpretation was conducted through SHapley Additive exPlanations (SHAP) analysis. Subsequently, nomograms were developed to estimate the risk of MDR-KP infection in hospitalized individuals. Finally, the association between independent variables and the risk of MDR-KP acquisition was elucidated through Restricted Cubic Spline (RCS) analysis.
RESULTS: The results of the multivariable logistic regression analysis indicated that C-reactive protein (CRP), Uric Acid (UA), Urea, Platelet (PLT), Hemoglobin (HB), Red blood cell counts (RBC), Age, and Gender were identified as independent risk factors. We incorporated these independent risk factors into six machine learning, and found that the XGBoost-based model exhibited superior performance compared to other machine learning algorithms, achieving a recall of 0.732, an F1 score of 0.707, and an AUC of 0.777. Furthermore, the SHAP method highlighted Urea, UA, and PLT as the primary decision factors predicted by the machine learning model, and the RCS analysis revealed a nonlinear relationship between Age, CRP, RBC, HB, UA, UREA, and the risk of MDR-KP infection.
CONCLUSION: This study has developed an effective risk prediction model for MDR-KP infection. The model has the potential to assist healthcare providers in early identification of high-risk patients, enabling timely implementation of preventive and therapeutic interventions.