基于交互式网页计算器的重症监护病房28天内输血的预测模型建立与验证  

Construction and validation of prediction models of blood transfusion within 28 days in intensive care unit based on interactive online calculator

作　　者：盛松 高洪阳 张艳虹[1] 黄烨[1] 王幸 Sheng Song;Gao Hong-yang;Zhang Yan-hong;Huang Ye;Wang Xing(Emergency Department,Xiyuan Hospital,China Academy of Chinese Medical Science,Beijing 100091,China)

机构地区：[1]中国中医科学院西苑医院急诊科,北京100091

出　　处：《中国急救医学》2022年第4期307-312,共6页Chinese Journal of Critical Care Medicine

基　　金：2019国家中医药管理局中医药循证能力建设项目(ZZ13-024-4)。

摘　　要：目的基于交互式网页计算器建立重症监护病房(ICU)28 d内输血的预测模型并进行内部验证。方法前瞻性连续收集2016年1月至2017年9月在日本千叶大学附属医院外科/内科ICU收治的符合纳入排除标准的患者121例。原始研究收集了性别、年龄、体重指数(BMI)、ICU入院途径、序贯器官衰竭评分(SOFA)和脓毒症诊断等资料,以及ICU入院日采集的12项贫血相关生化指标和ICU住院期间输血前累积采血量。数据按4∶1比例随机拆分成建模组和验证组,使用Lasso回归和共线性筛查筛选建模变量,应用Logistic回归和R shiny构建28 d输血预测模型和交互式网页计算器,然后从模型校准度(hosmer-lemeshow检验)、区分度(受试者工作特性曲线下面积)和有效性(决策曲线分析)三方面评价建模组和验证组模型性能。结果建模变量为SOFA评分、转铁蛋白(TRF)和累积采血量,模型Logit(Y)=-1.54480+0.21694×SOFA评分-0.02274×TRF(mg/dL)+0.00707×累积采血量(mL),基于R shiny建立的交互式网页计算器访问网址为https://shengsong.shinyapps.io/BLOOD_TRANSFUSION。建模组和验证组模型经校准度、区分度和有效性评价性能较好。结论本研究建立的预测模型可以优化ICU危重症患者输血策略,并且具有良好的预测性能。Objective To build and internally validate prediction models of blood transfusion within 28 days in intensive care unit(ICU)based on interactive online calculator.Methods Our study prospectively collected the patients in surgical/medical ICU of Chiba University Hospital between January 2016 and September 2017 and a total of 121 patients who met the inclusion and exclusion criteria were included in the study.Demographic data and clinical information including age,gender,body mass index(BMI),ICU admission route,sequential organ failure assessment(SOFA)and sepsis diagnosis were collected.Also,12 anemia-related biochemical indicators on ICU admission and cumulative amount of exsanguination before blood transfusion during ICU stay were recorded.Patients were randomly split into development group and validation group at the ratio of 4:1.Later,Lasso regression and collinearity analysis were used to screen the variables for multivariable modelling.Multivariable Logistic regression and R shiny package were used to build prediction models for blood transfusion within 28 days and interactive online calculator,respectively.The performance of prediction model and internal validation were evaluated from the aspects of calibration(hosmer-lemeshow test),discrimination(the area under the receiver operating characteristic curve)and effectiveness(decision curve analysis).Results Variables for multivariable modelling were SOFA,transferrin(TRF)and cumulative amount of exsanguination.The obtained formula of the model was:Logit(Y)=-1.54480+0.21694×SOFA-0.02274×TRF(mg/dL)+0.00707×cumulative amount of exsanguination(mL).The interactive online calculator based on R shiny was accessible at https://shengsong.shinyapps.io/BLOOD_TRANSFUSION.The performance of calibration,discrimination and effectiveness were excellent in both development group and validation group.Conclusions The prediction models with excellent performance can optimize blood transfusion strategies for critically ill patients in ICU.

关 键 词：交互式网页计算器 R shiny 重症监护病房(ICU) 输血 预测模型 

分 类 号：R457.1[医药卫生—治疗学]