|
Abstract:
|
In multivariate survival analysis, copulas have become a popular tool for modeling the joint survival function using a vector of continuous time-to-event random variables. However, in the presence of right censoring, selecting a parametric copula is a non-trivial task that may result in model misspeci cation. To avoid fitting a misspeci ed parametric model, and develop a goodness-of- t test for a general copula, nonparametric copula estimation can play an important role. To achieve this goal, we use empirical likelihood to develop nonparametric estimators of a copula function for censored multivariate time-to-event data. The asymptotic properties of the logarithmic empirical likelihood (EL) ratio statistic are derived, and the finite sample properties of the EL-based estimator are assessed via a simulation study. The proposed method is illustrated using the data from 197 high-risk patients who were participants in the Diabetic Retinopathy Study (DRS).
|
