资源描述：
《急性机械性绞窄性肠梗阻预测模型的建立和检验.pdf》由会员上传分享，免费在线阅读，更多相关内容在行业资料-天天文库。
1、北京医科大学学报JOURNALOFBEIJINGMEDICALUNIVERSITYVol.32No.2Apr.2000183急性机械性绞窄性肠梗阻预测模型的建立和检验12341551王立新,史宗道,赵一鸣,汪欣,蒋斌,顾晋,姜可伟,周孝思(1北京医科大学第三医院普通外科,北京100083;3临床流行病学中心;4第一医院普通外科;5人民医院普通外科;2华西医科大学临床流行病学教研室)[关键词]肠梗阻;预测;回归分析;疾病模型,动物[摘要]目的:应用多变量分析方法建立急性机械性绞窄性肠梗阻
2、的预测模型。方法:总结3家医院7至10年间收治的急性机械性肠梗阻571例,分别应用logistic回归和分类与回归树(classificationandregressiontree,CART)方法建立急性绞窄性肠梗阻的预测模型,用交叉应证方法对模型进行检验,并与临床经验的诊断进行比较。结果:腹痛性质、压痛包块、腹水征、肌紧张、既往手术史和类似发作史是建立模型的重要变量,其中前4项与肠绞窄呈正相关,后两项与肠绞窄呈负相关;logistic模型的敏感度为71.8%,特异度为73.1%,CART模型的敏感度为73.
3、1%,特异度为69.8%,两种模型的敏感度明显高于临床医师的经验判断(45.2%)。结论:logistic和CART模型对急性机械性绞窄性肠梗阻具有预测作用,对临床诊断有参考意义。[中图分类号]R574.2332[文献标识码]A[文章编号]10001530(2000)02018304Developmentandtestingofpredictivemodelforacutemechanicalstrangulatedintestinalobstruction1234WANGLiXin,SHI
4、ZongDao,ZHAOYiMing,WANGXin,1551JIANGBin,GUJin,JIANGKeWei,ZHOUXiaoSi[1DepartmentofSurgery,theThirdHospital,BeijingMedicalUniversity(BMU),Beijing100083,China;2.DepartmentofClinicalEpidemiology,WestChinaUniversityofMedicalSciences;3.CenterofClinicalEpide
5、miology,theThirdHospital,BMU;4.DepartmentofSurgery,theFirstHospital,BMU;5.DepartmentofSurgery,thePeoplesHospital,BMU]KEYWORDSIntestinalobstruction;Forecasting;Regressionanalysis;Diseasemodels,animalABSTRACTObjective:Todeveloppredictivemodelsthatcandistin
6、guishacutestrangulationfromsimpleintestinalobstructionbymultivariateanalysis.Methods:Allthe571patientsinvolvedinthestudywerethecasesadmittedwithacutemechanicalintestinalobstructioninthedepartmentsofgeneralsurgeryofthreehospitalsaffiliatedtoBeijingMedicalUn
7、iversityinthepast7to10years.Logisticregressionmodelwasdevelopedandanalgorithmfortheclassificationofintestinalobstructionwasconstructedbyclassificationandregressiontrees(CART)analysis.Thesetwomodelsweretestedbythemethodofcrossvalidationandcomparedwiththes
8、urgeonsexperience.Results:Thepatternofabdominalpain,tendermass,ascites,abdominalrigidity,historyofabdominaloperationandhistoryofsimilarattackwereimportantvariablestodevelopthemodels;Sensitivity(SEN)o
