Statistical Inference for Competing Risks Model using Censored Data

Summary: In my invited talk, I will present recent work on analyzing a competing risks model under a unified hybrid censoring scheme. The study assumes Weibull-distributed latent failure times with a common shape and distinct scale parameters. I will discuss the derivation of maximum likelihood estimates, conditions for their uniqueness, and the construction of approximate confidence intervals. The talk will also cover Bayesian estimation techniques under both symmetric and asymmetric loss functions, incorporating informative and non-informative priors. Using Markov Chain Monte Carlo with Gibbs sampling, we obtain approximate Bayes estimates and highest posterior density intervals, while also evaluating coverage probabilities. Furthermore, I will highlight the use of likelihood ratio tests for hypothesis testing, supported by Monte Carlo simulations that compare the performance of the estimators. To conclude, I will demonstrate the practical utility of the proposed methods through the analysis of one real-life data set.

Link para acesso: Statistical Inference for Competing Risks Model using Censored Data