混合效应随机微分方程的贝叶斯分析.pdf

《混合效应随机微分方程的贝叶斯分析.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、JournalofSoutheastUniversity(EnglishEdition)Vo1．30，No．1，PP．122—127Mar．2014ISSN1003—7985Bayesiananalysisformixed-effectsmodeldefinedbystochasticdifferentialequationsYanFangrong，，ZhangPing，LuTao，LinJinguan(DepartmentofMathematics，SoutheastUniversity，Nanjing210096，Chi

2、na)(DepartmentofMathematics，ChinaPharmaceuticalUniversity，Nanjing210009，China)(LaboratoryofMolecularDesignandDrugDiscovery，ChinaPharmaceuticalUniversity，Nanjing210009，China)(StateKeyLaboratoryofNaturalMedicines，ChinaPharmaceuticalUniversity，Nanjing210009，China)Abst

3、ract：Thenonlinearmixed—effectsmodelwithstochasticdeveloped．differentialequations(SDEs)isusedtomodelthepopulationNonlinearmixed．effectsmodeling(N0NMEM1haspharmacokinetic(PPK)datathatareextendedfromordinarybeenprovedtobeakindofwidelyusedandeffectivetooldifferentialeq

4、uations(0DEs)byaddingastochastictermtofordescribingthepharmacokinetic(PK)andpharmaco—thestateequation．Comparedwiththe0DEs．theSDEscandynamic(PD)propertiesofdrugs1-21．Theflexibilityofmodelcorrelatedresidualswhichareubiquitousiuactualmodelingcorrelationstructures，with

5、thetotalvariationpharmacokineticproblems．TheBayesianestimationisseparatedintointer．andintra．individualvariationcontrib．providedfornonlinearmixed—effectsmodelsbasedonutestotherapidlygrowingquantityofnonlinearmixed—stochasticdifferentialequations．CombiningtheGibbsand

6、theeffectsmodels．ThismodelingapproachusuallyleadstoMetropolis—Hastingsalgorithms，thepopulationandindividualmoreaccurateestimationofpopulationparameters，espeparametervaluesaregiventhroughtheparameterposteriorciallyinpharmacokinetic／pharmacodynamic(PI<／／PD)predictive

7、distributions．Theanalysisandsimulationresultsmodelsincludingrandomeffects．However．thetradition．showthattheperformanceoftheBayesianestimationforalnonlinearmixed—effectsmodelsaremostfrequentlymixed—effectsSDEsmodelandanalysisofpopulationbasedonordinarydifferentialequ

8、ations(ODEs)，assum．pharmacokineticdataisreliable．Theresultssuggestthattheingthattheinter—individualvariationisindependentwimproposedmethodisfeasi