重复测量数据的混合模型及其MIX_省略_线性模型及其SAS软件实现_二_张岩波

＊(20001019)MIXED＊———混合线性模型及其SAS软件实现(二)(030001)　张岩波　何大卫　刘桂芬　张晋昕　郭　静　　【　】　　,,SAS。　MIXED。　,。　MIXED、。【】　　　　MIXED　　(Repeatedmeasuresdata)(Longitudeda-ta),、。,。,。。,〔1〕,,,,,,。、。SASGLMREP。,GoldsteinMLN/MLwiN〔1,2〕。SAS,MIXED,,,MIXED。GLM、、〔3〕。,。,?,A、B,:y=μ+α+u+β+(αβ)+eμ,α、β、(αβ)A、B,u、e。B,,,。,,(、)。,,。,。,,MIXED,ESTIMATED、LSMEANS、RANDOM、REPEATED。24(G)、PC、AD.PAD.T(A),HR、5、15、30、60(T),1。1　HR5′15′30′60′11251261201341342115108118128118::::::6115126120126122PC11201201201151202137127117127114::::::6135110114122110AD.P11351081071081182125109126126128::::::6134108108108104AD.T11251021081171182125106110116136::::::612410298108108·272·　200110185　　,〔4〕。,,,,。MIXED。Huynh-Feldt,RE-PEATED,TYPE=HFTYPE=UN(Unstructured),,:Chi-Squre=43.7443,ν=9P0.0001(15;HF,6,9),Huynh-Feldt,。,2。2　TheMIXEDProcedureUnstructuredCovariance(ParameterEstimates)matrixRowCOL1COL2COL3COL4COL51177.01＊＊-84.56＊＊-81.44＊＊-49.33-67.622-84.56＊＊101.55＊76.04＊22.5850.663-81.44＊76.04＊97.18＊＊26.1860.334-49.3322.5826.1851.20＊＊36.835-67.6250.6660.3336.8387.39＊ModelFittingInformationforHRDescriptionValueObservations120.0000VarianceEstimate1.0000StandardDeviationEstimate1.0000LogLikelihood-412.753Akaike'sInformationCriterion-427.753Schwarz'sBayesianCriterion-448.660-2LogLikelihood825.5068NullModelLRTChi-Square57.4302NullModelLRTDF14.0000NullModelLRTP-Value0.0000SolutionforFixedEffectsParameterEstimateStdErrorDDFTPINTERCEPT114.025601311.626029612070.130.0001A19.643650692.29955312204.190.0004A29.971566662.29955312204.340.0003A33.946332492.29955312201.720.1016T0.226819630.06040416203.760.0012T(A)1-0.236631180.0854243820-2.770.0118T(A)2-0.316240880.0854243820-3.700.0014T(A)3-0.132273890.0854243820-1.550.1372TestsofFixedEffectsSourceNDFDDFTypeⅢFPA3208.730.0007T1203.380.0809T(A)3205.130.0086　　＊P0.05　＊＊P0.01　　2、、。(NullModel)(LRT)Chi-Squre=57.43,v=14,P=0.000。14151。:HR。(A)(A1、A2、A3、A4),TA2、。3。　　3,。,A2,〔4〕。,TA,():modelhr=ata＊t/s;A、T。　　,MIXED:·273·　ChineseJournalofHealthStatistics,October2001,Vol.18,No.53　TheMIXEDProcedureModelFittingInformationforHRDescriptionValueObservations120.0000VarianceEstimate49.8312StandardDeviationEstimate7.0591LogLikelihood-411.920Akaike'sInformationCriterion-415.920Schwarz'sBayesianCriterion-421.495-2LogLikelihood823.8392NullModelLRTChi-Square9.3523NullModelLRTDF3.0000NullModelLRTP-Value0.0250SolutionforFixedEffects()ParameterEstimateStdErrorDDFTPINTERCEPT132.721107831.953100488467.950.0001A2-4.050479842.135045715-1.900.1163A3-4.778017592.1530147484-2.220.0292A4-7.136019522.1530147484-3.310.0014T-5.405479800.7387602923-7.320.0001T＊T0.474088380.08338129235.690.0001T＊T＊T-0.013884740.0028576923-4.860.0001T＊T＊T＊T0.000123920.00002782234.450.0002CovarianceParameterEstimates(MLE)()CovParmRatioEstimateStdErrorZPTUN(1,1)0.000263770.013144210.007409121.770.0761UN(2,1)-0.02060673-1.026857850.50630056-2.030.0425UN(2,2)1.0118762750.4229923237.232374791.350.1756Residual1.0000000049.831183907.379446066.750.0001　　modelhr=ata＊t;randomgt＊g;randominterceptt/sub=g;GLMTEST,。,TYPE=UNHFHuynh-Feldt。VC(Variancecompents)、AR(1)。GLM,SUBTYPE,REPEATED。LSMEANS、ESTIMATE、CON-TRASTGLM,,MIXED,。,ML,REML。,Z,-2LogLikelihood。〔4〕,MIXEDMLN。MIXED1E-08,。MLN,(1E-04)。,。MIXED、。MixedModelforRepeatedMeasuredDataandApproachofProcMixed———MixedLinearModelanditsImplementationsofSAS(2).　Zhangyanbo,HeDawei,LiuGuifenetal.,Dept.ofhealthStatistics,ShanxiMedicaluniversity(030001),Taiyuan　　【Abstract】　Objective　Becauseofcorrelationamongrepeatedmeasuresdataanditshierarchicalerror,itisnotsuitabletouseordinalmethod.Methods　AexampleofrepeatedmeasurementwasanalyzedwithmixedmodelusingMIXEDprocedure.Results　Repeatedmeasurementobtainedreasonableresultsbythefixedandrandomeffectsalongwitheffi-cientestimateofcovariancematrix.Conclusion　Mixedmodelcaneffec-tivelyanalyzerepeatedmeasuresdatawithSASMIXED.【Keywords】　Repeatedmeasuresdata　Mixedlinearmodel　Multilevelmodel　Mixedprocedure　　　1.,,.REP.,1998,15(6):5-8.2.G.Woodhouse.MultilevelModellingApplicationsAguideforusersofMln.InstituteofEducation,UniversityofLondon,1996,9-57.·274·　2001101853.SASInstitute.SAS/STATSoftware:ChangesandEnhancements.Re-lease6.11,1996.4.,..,1997,24(1):12-14.:datarepe;inputagthr;a1=0;a2=0;a3=0;a4=0;ifa=1thena1=1;ifa=2thena2=1;ifa=3thena3=1;ifa=4thena4=1;output;datalines;110125115126::::run;procmixeddata=repemethod=mlmaxiter=50;　classag;　modelhr=/s;　repeated/type=unsub=gr;run;procmixeddata=repemethod=remlmaxiter=50;　classg;　modelhr=/s;　repeated/type=hfsub=gr;run;procmixeddata=repemethod=mlmaxiter=50;　classag;　modelhr=att(a)/s;　repeated/type=unsub=gr;run;procmixeddata=repemethod=mlmaxiter=50;　classg;　modelhr=a2a3a4tt＊tt＊t＊tt＊t＊t＊t/s;　randomta2/type=unsub=gg;run;浅谈出院病案首页在住院统计工作中的重要性(621000)　肖　强　赵　燕　　,,,。,,。,、,,。、,。,,,。,,、、。,,、,。,、、、、、,,,(、、)。、、、、,,、、,。、,,,。,,。。,。,,,。、,1～3,,,6。,,。,、。,、、,,,。,。,,,。,。,、、,。,,、。、,,。,。,、,,,、、。,,。。,,、,,,。`',。,。　　　1.,..:,1992,373-408.·275·　ChineseJournalofHealthStatistics,October2001,Vol.18,No.5