基于ROC曲线的儿科冬季床位超负荷预警时机研究及管理验证OA

Objective Aiming at the problem of lack of quantitative basis for intervention timing in the allocation of pediatric peak beds in winter,a data-driven intervention timing prediction method was explored and its management efficiency was verified.Methods Taking Tianjin Children's Hospital as the research object,a two-stage design was adopted.In the first stage(2023),receiver operating characteristic(ROC)curve analysis was used to define the overload state with the bed efficiency index and determine the optimal intervention lead time.In the second stage(2024),the adjustment of"surgical integration and internal medicine expansion"was implemented according to this opportunity,and compared with the empirical adjustment in 2023.Wilcoxon rank sum test and binary Logistic regression were used to evaluate the effect and control confounding.Results The ROC curve determined that the best intervention advance period was 42 days before the peak(area under the curve was 0.780).After the model-guided adjustment in 2024,the bed overload in all wards was significantly alleviated,the average length of stay was shortened,and the average cost per time was reduced by 19.5%,which were significantly better than those after the empirical adjustment in 2023(P<0.001).Binary Logistic regression analysis confirmed that the early intervention period was an independent protective factor for the bed overload state(OR=0.973,P=0.013).The subgroup analysis suggested that the respiratory department was suitable for intervention 48 days in advance,and the digestive department was suitable for dynamic response.Conclusion The pediatric bed overload early warning model constructed with the ROC curve provides empirical support for the improvement of pediatric service capabilities.The hierarchical response strategy has achieved a transformation from empirical judgment to data-driven,providing a quantitative decision-making tool for seasonal peak response.