R语言生存分析大全.docx

R语言生存分析大全.docx

《R语言生存分析大全.docx》由会员分享，可在线阅读，更多相关《R语言生存分析大全.docx（18页珍藏版）》请在冰豆网上搜索。

R语言生存分析大全

CRANTaskView:

SurvivalAnalysis

Maintainer:

ArthurAllignolandAurelienLatouche

Contact:

arthur.allignolatuni-ulm.de

Version:

2016-01-27

Survivalanalysis,alsocalledeventhistoryanalysisinsocialscience,orreliabilityanalysisinengineering,dealswithtimeuntiloccurrenceofaneventofinterest.However,thisfailuretimemaynotbeobservedwithintherelevanttimeperiod,producingso-calledcensoredobservations.

ThistaskviewaimsatpresentingtheusefulRpackagesfortheanalysisoftimetoeventdata.

Pleaseletthe maintainers knowifsomethingisinaccurateormissing.

StandardSurvivalAnalysis

EstimationoftheSurvivalDistribution

∙Kaplan-Meier:

 The survfit functionfromthe survival packagecomputestheKaplan-Meierestimatorfortruncatedand/orcensoreddata. rms （replacementoftheDesignpackage）proposesamodifiedversionofthe survfit function.The prodlim packageimplementsafastalgorithmandsomefeaturesnotincludedin survival.VariousconfidenceintervalsandconfidencebandsfortheKaplan-Meierestimatorareimplementedinthe km.ci package. plot.Surv ofpackage eha plotstheKaplan-Meierestimator.The NADA packageincludesafunctiontocomputetheKaplan-Meierestimatorforleft-censoreddata. svykm in survey providesaweightedKaplan-Meierestimator. nested.km in NestedCohort estimatesthesurvivalcurveforeachlevelofcategoricalvariableswithmissingdata.The kaplan-meier functionin spatstat computestheKaplan-Meierestimatorfromhistogramdata.The MAMSE packagepermitstocomputeaweightedKaplan-Meierestimate.The KM functioninpackage rhosp plotsthesurvivalfunctionusingavariantoftheKaplan-Meierestimatorinahospitalisationriskcontext.The survPresmooth packagecomputespresmoothedestimatesofthemainquantitiesusedforright-censoreddata,i.e.,survival,hazardanddensityfunctions.The asbio packagepermitstocomputetheKaplan-MeierestimatorfollowingPollocketal.（1998）.The bpcp packageprovidesseveralfunctionsforcomputingconfidenceintervalsofthesurvivaldistribution（e.g.,betaproductconfidenceprocedure）.The lbiassurv packageoffersvariouslength-biascorrectionstosurvivalcurveestimation.Non-ParametricconfidancebandsfortheKaplan-Meierestimatorcanbecomputedusingthe kmconfband package.The kmc packageimplementstheKaplan-Meierestimatorwithconstraints.

∙Non-Parametricmaximumlikelihoodestimation（NPMLE）:

 The Icens packageprovidesseveralwaystocomputetheNPMLEofthesurvivaldistributionforvariouscensoringandtruncationschemes.MLEcens canalsobeusedtocomputetheMLEforinterval-censoreddata. dblcens permitstocomputetheNPMLEofthecumulativedistributionfunctionforleft-andright-censoreddata.The icfitfunctioninpackage interval computestheNPMLEforinterval-censoreddata.The DTDA implementsseveralalgorithmspermittingtoanalysepossiblydoublytruncatedsurvivaldata.

∙Parametric:

 The fitdistrplus packagepermitstofitanunivariatedistributionbymaximumlikelihood.Datacanbeintervalcensored.The vitality packageprovidesroutinesforfittingmodelsinthevitalityfamilyofmortalitymodels.

HazardEstimation

∙The muhaz packagepermitstoestimatethehazardfunctionthroughkernelmethodsforright-censoreddata.

∙The epi.insthaz functionfrom epiR computestheinstantaneoushazardfromtheKaplan-Meierestimator.

∙polspline, gss and logspline allowtoestimatethehazardfunctionusingsplines.

∙The ICE packageaimsatestimatingthehazardfunctionforintervalcensoreddata.

∙The bshazard packageprovidesnon-parametricsmoothingofthehazardthroughB-splines.

Testing

∙The survdiff functionin survival comparessurvivalcurvesusingtheFleming-HarringtonG-rhofamilyoftest. NADA implementsthisclassoftestsforleft-censoreddata.

∙clinfun implementsapermutationversionofthelogranktestandaversionofthelogrankthatadjustsforcovariates.

∙The exactRankTests implementstheshift-algorithmbyStreitbergandRoehmelforcomputingexactconditionalp-valuesandquantiles,possiblyforcensoreddata.

∙SurvTest inthe coin packageimplementsthelogranktestreformulatedasalinearranktest.

∙The maxstat packageperformstestsusingmaximallyselectedrankstatistics.

∙The interval packageimplementslogrankandWilcoxontypetestsforinterval-censoreddata.

∙Threegeneralisedlogranktestsandascoretestforinterval-censoreddataareimplementedinthe glrt package.

∙survcomp compares2hazardratios.

∙The TSHRC implementsatwostageprocedureforcomparinghazardfunctions.

∙The Survgini packageproposestotesttheequalityoftwosurvivaldistributionsbasedontheGiniindex.

∙The FHtest packageoffersseveraltestsbasedontheFleming-Harringtonclassforcomparingsurivalcurveswithright-andinterval-censoreddata.

∙The LogrankA packageprovidesalogranktestforwhichaggregateddatacanbeusedasinput.

∙Theshorttermandlongtermhazardratiomodelfortwosamplessurvivaldatacanbefoundinthe YPmodel package.

RegressionModelling

∙Coxmodel:

 The coxph functioninthe survival packagefitstheCoxmodel. cph inthe rms packageandthe eha packageproposesomeextensionstothe coxph function.Thepackage coxphf implementstheFirth'spenalisedmaximumlikelihoodbiasreductionmethodfortheCoxmodel.AnimplementationofweightedestimationinCoxregressioncanbefoundin coxphw.The coxrobust packageproposesarobustimplementationoftheCoxmodel. timecox inpackage timereg fitsCoxmodelswithpossiblytime-varyingeffects.The mfp packagepermitstofitCoxmodelswithmultiplefractionalpolynomial.The NestedCohort fitsCoxmodelsforcovariateswithmissingdata.ACoxmodelmodelcanbefittedtodatafromcomplexsurveydesignusingthe svycoxph functionin survey.The multipleNCC packagefitsCoxmodelsusingaweightedpartiallikelihoodfornestedcase-controlstudies.The intcox packageimplementstheCoxmodelforinterval-censoreddatausingtheICM-algorithm.The MIICD packageimplementsPan's（2000）multipleimputationapproachtoCoxmodelsforintervalcensoreddata.The ICsurv packagefitsCoxmodelsforinterval-censoreddatathroughanEMalgorithm.The dynsurvpackagefitstime-varyingcoefficientmodelsforintervalcensoredandrightcensoredsurvivaldatausingaBayesianCoxmodel,asplinebasedCoxmodeloratransformationmodel.The CPHshapepackagecomputestheCoxproportionalhazardsmodelwithshapeconstrainedhazardfunctions.The OrdFacReg packageimplementstheCoxmodelusinganactivesetalgorithmfordummyvariablesoforderedfactors.The survivalMPL packagefitsCoxmodelsusingmaximumpenalisedlikelihoodandprovideanonparametricsmoothestimateofthebaselinehazardfunction. 

The cumres functionin gof computesgoodness-of-fitmethodsfortheCoxproportionalhazardsmodel.Theproportionalityassumptioncanbecheckedusingthe cox.zph functionin survival.The CPEpackagecalculatesconcordanceprobabilityestimatefortheCoxmodel,asdoesthe coxphCPE functionin clinfun.The coxphQuantile inthelatterpackagedrawsaquantilecurveofthesurvivaldistributionasafunctionofcovariates.The multcomp packagecomputessimultaneoustestsandconfidenceintervalsfortheCoxmodelandotherparametricsurvivalmodels.The lsmeans packagepermitstoobtainleast-squaresmeans（andcontraststhereof）fromlinearmodels.Inparticular,itprovidessupportforthe coxph, survreg and coxme functions.The multtest packageonBioconductorproposesaresamplingbasedmultiplehypothesistestingthatcanbeappliedtotheCoxmodel.TestingcoefficientsofCoxregressionmodelsusingaWaldtestwithasandwichestimatorofvariancecanbedoneusingthe sawspackage.The rankhazard packagepermitstoplotvisualisationoftherelativeimportanceofcovariatesinaproportionalhazardsmodel.The smoothHR packageprovideshazardratiocurvesthatallowsfornonlinearrelationshipbetweenpredictorandsurvival.The paf packagepermitstocomputetheunadjusted/adjustedattributablefractionfunctionfromaCoxproportionalhazardsmodel.The PHevalpackageproposestoolstochecktheproportionalhazardsassumptionusingastandardisedscoreprocess.

∙ParametricProportionalHazardsModel:

 survreg （from survival）fitsaparametricproportionalhazardsmodel.The eha and mixPHM packagesimplementaproportionalhazardsmodelwithaparametricbaselinehazard.The pphsm in rms translatesanAFTmodeltoaproportionalhazardsform.The polspline packageincludesthe hare functionthatfitsahazardregressionmodel,usingsplinestomodelthebaselinehazard.Hazardscanbe,butnotnecessarily,proportional.The flexsurv packageimplementsthemodelofRoystonandParmar（2002）.Themodelusesnaturalcubicsplinesforthebaselinesurvivalfunction,andproportionalhazards,proportionaloddsorprobitfunctionsforregression.

∙AcceleratedFailureTime（AFT）Models:

 The survreg functioninpackage survival canfitanacceleratedfailuretimemodel.Amodifiedversionof survreg isimplementedinthe rms package（ psmfunction）.Itpermitstousesomeofthe rms functionalities.The eha packagealsoproposesanimplementationoftheAFTmodel（function aftreg）.AnAFTmodelwithanerrordistributionassumedtobeamixtureofG-splinesisimplementedinthe smoothSurv package.The NADA packageproposesthefrontendofthe survreg functionforleft-censoreddata.Aleast-squareprincipledimplementationoftheAFTmodelcanbefoundinthe lss package.The simexaft packageimplementstheSimulation-ExtrapolationalgorithmfortheAFTmodel,thatcanbeusedwhencovariatesaresubjecttomeasurementerror.Arobustversionoftheacceleratedfailuretimemodelcanbef