WORKSHOP ON DEPENDENCE ANALYSIS
26 and 27 of SEPTEMBER, 2025
Institute of Mathematics, Statistics and Scientific
Computation of the University of Campinas, Brazil
SATELLITE WORKSHOP OF
8th BRAZILIAN CONFERENCE ON STATISTICAL MODELING IN INSURANCE AND FINANCE
Title: A Shared Frailty Regression Model for Clustered Survival Data
In this talk, I will present a new frailty model for multivariate lifetimes based on a mixture of inverse Gaussian distributions. The model distinguishes itself by determining the mixture weights through direct parameterization, with its closed-form Laplace transform enabling a straightforward assessment of dependence using Kendall's tau. Both the parametric and semiparametric versions of the model are discussed, along with the advantages of using the EM algorithm for robust parameter estimation. Results from Monte Carlo simulations and applications to two cancer datasets are presented, highlighting the benefits of this approach over traditional frailty models. Finally, the methodology is shown to be readily accessible via an R package.