Bootstrapping Regression Parameters in Multivariate Survival Analysis

摘要: Bootstrap methods are proposed for estimating sampling distributions and associated statistics regression parameters in multivariate survival data. We use an Independence Working Model (IWM) approach, fitting margins independently, to obtain consistent estimates of the marginal models. Resampling procedures, however, applied appropriate joint distribution estimate covariance matrices, make bias corrections, construct confidence intervals. The allow fixed or random explanatory variables, latter case using extensions existing resampling schemes (Loughin,1995), they permit possibility censoring. An application is shown viral positivity time data previously analyzed by Wei, Lin, Weissfeld (1989). A simulation study small-sample properties shows that bootstrap procedures provide substantial improvements variance estimation over robust estimator commonly used with IWM.