新药临床试验中重复测量资料的混合效应模型-陈峰

整理文档很辛苦,赏杯茶钱您下走!

免费阅读已结束,点击下载阅读编辑剩下 ...

阅读已结束,您可以下载文档离线阅读编辑

资源描述

*(39500127)1.(226001)2.3.4.5.··陈 峰1 姚 晨2 孙 高3 任仕泉4 何清波5 苏炳华5 陆守曾1  Ⅱ,、,。,〔1〕。。,,:t,“”(,)t。,:yij=α+β1treat+β2yi0+β3time+eij(1),i=1,…,n,j=1,2,…,k,j=0。yijij,yi0i,(baseline)。treat,treat=0,treat=1。time,time=0。β2yi0;,,β2yi0。β3time(time,(dummyvariables))。eij,,N(0,σ20)。,,(power),;,(in-tra-subjectcorrelation),Ⅰ,。k,k,k,,。,nk,。,。(mixedmodel)〔2〕。,。:、、、〔3〕。。1. ,(exchangeable),(Compoundsymmetry)。,ρ。2. 。(5):R=1ρ1000ρ11ρ2000ρ21ρ3000ρ31ρ4000ρ41(2),1(sta-tionary1-dependence),1(nonstationary1-dependence)。2:R=1ρρ00ρ1ρρ0ρρ1ρρ0ρρ1ρ00ρρ1(3)·373· 200012176k。,,()。3.(auto-correlation) :ρk=ρ|j-k|(4)ρ,,。1(firstorderauto-correlation)。,:ρjk=ρd|tj-tk|(5)。d。4.(unstructured,generalstructure) 。5.(independent,zerocorrelation) :“AB”BA,AB。,,。:yij=αi+β1xij+…+βmxij(6),αii,:αi=α+ui(7)α,uii,∑ui=0,。ui0,。,uiN(0,σ21),:ρ=σ21σ21+σ20(8),。,yij(,Poisson),:yij=μijg(μij)=αi+β1xij+…+βmxij(9),g(·)(linkfunction)。yij,g(·)logitProbit;yijPoisson,g(·)。1() ,72,2∶12,,,(MDNS)。MDNS,28,2。1。,treat(0,1),center3,。1 72treatCenterMDNS71421286090111282014121214142012622241481212…………………………711326222221141012721315141312121310  ,,。,:yij=β1time+β2treat·time+β3ts+β4treat·ts+β5c2+β6c3+eij,time;treat(treat=1)(treat=0);stage(stage=0)(stage=1),tstime·stage;c1,c2。2。2 zPtime0.33380.023514.230.0001treat·time0.07290.03182.290.0225time·stage-0.19560.02228.790.0001treat·ts0.02640.03390.780.4370c1-4.25230.5400-7.870.0001c2-5.12020.53869.510.00012,MDNS;MDNS,。。:yij=ui+β1time+β2treat·time+β3ts+β4treat·ts+β5c2+β6c3+eij,MDNS。,。:7 14 21 28 60 901  0.91251  0.86230.93801  0.83320.91280.91771  0.57550.65750.69870.70311  0.49050.59350.59290.67990.88701 ·374· ChineseJournalofHealthStatistics,December2000,Vol.17,No.6  ,,,(,63)。,。(4)(5),(5),281。-1102.18-1096.99。(5)。:7 14 21 28 60 901  0.87471  0.76510.87471  0.66920.76510.87471  0.39180.44790.51200.58541  0.22930.26220.29970.34270.58541   1(28)0.5854。3。3 1zPtime0.15750.03324.750.0001treat·time0.06360.03931.620.1069time·stage-0.08840.02982.970.0032treat·ts-0.01610.03630.440.6580c1-2.85350.97322.930.0046c2-2.87980.96262.990.0038  3,MDNS,。。,MDNS,281.7808;904.275。。,。2((countdata)) ,59。8,。2,,,,8〔8〕。4。4 59treatoffstimeagey1080311110213151022313………………5912337359124372,treat(0,1),offs,time(0,1,2,3,48),age,y。(2),,Poisson。ln(λij)=αi+β1treat+β2time+β3treat·time+β4age+β5yi0,yi0,。,0.4063,Deviance=950.45;,0.5064t,t(2),Deviance=951.73。,,Deviance。。5。5 2zPtreat-0.10840.10051.0790.281time-0.04150.02221.8680.062time·treat-0.03150.03131.0060.315age0.01890.00503.8180.000y00.02270.000829.9530.0000.77690.18034.3090.000  ,,,,;。,。,0.7950,Deviance=2495.89。,,0.7757,Deviance=3538.35。,。  1.,、,。,,Iα,1-β,:N0=(uα+uβ)2σ2δ2,,k。(ρ),Iα,1-β,:N=nk=〔1+(k-1)ρ〕N0,n,N=nk。,n·375· 200012176N0;nkN0。,nk:nk1+(k-1)ρ。,,nk。,t,n,。,,。1,,:N=〔1+2(k-1)kρ+2(k-2)kρ2+…+2kρk-1〕N0  1,,:N=1+2(k-1)kρN0  ,。2.,,。bootstrap,〔5,6〕。3.,,,,。,。4.,,,;,。5.。(1)、,SAS;(2),,、、,Stata(V5.0);(3),,,,、,,,MLnMLwiN()。   1.,,..,1998,25(3):269-271.2.RPrescott,HBrown.MaxedmodelsanalysisofclinicaltrialsusingSAS:procmixedandbeyond.CoursePreseters.1995.3.,..,1998,15(4):4-8.4.DigglePJ,LiangKY,ZegerST.Analysisoflongitudinaldata.Oxford:ClarendonPress.1995,13-14.5.,,..,1999,19(6):359-362.6.,,.Bootstrap.,1997,14(5):5-7.医院统计工作应遵循的一般原则(272111) 宗珊珊  ,。、,、、,、、。。,。、,。,,,。、,。,,。,。,,,,。·376· ChineseJournalofHealthStatistics,December2000,Vol.17,No.6

1 / 4
下载文档,编辑使用

©2015-2020 m.777doc.com 三七文档.

备案号:鲁ICP备2024069028号-1 客服联系 QQ:2149211541

×
保存成功