We study nonparametric estimation of the sub-distribution func-tions for current status data with competing risks. Our main interest is in the nonparametric maximum likelihood estimator (MLE), and for comparison we also consider a simpler ‘naive estimator
CURRENTSTATUSDATAWITHCOMPETINGRISKS:
CONSISTENCYANDRATESOFCONVERGENCEOFTHEMLE
ByPietGroeneboom,MarloesH.Maathuis
andJonA.Wellner
DelftUniversityofTechnologyandVrijeUniversiteitAmsterdam,
UniversityofWashingtonandUniversityofWashington
Westudynonparametricestimationofthesub-distributionfunc-
tionsforcurrentstatusdatawithcompetingrisks.Ourmaininterest
isinthenonparametricmaximumlikelihoodestimator(MLE),and
forcomparisonwealsoconsiderasimpler‘naiveestimator’.Both
typesofestimatorswerestudiedbyJewell,VanderLaanandHenne-
man[7],butlittlewasknownabouttheirlargesampleproperties.We
havestartedto llthisgap,byprovingthattheestimatorsareconsis-
tentandconvergegloballyandlocallyatraten1/3.Wealsoshowthat
thislocalrateofconvergenceisoptimalinaminimaxsense.Theproof
ofthelocalrateofconvergenceoftheMLEusesnewmethods,and
reliesonarateresultforthesumoftheMLEsofthesub-distribution
functionswhichholdsuniformlyona xedneighborhoodofapoint.
OurresultsareusedinGroeneboom,MaathuisandWellner[3]to
obtainthelocallimitingdistributionsoftheestimators.
热门文档