周洁 副教授，首都师范大学

In this article, we propose a new joint modeling approach for recurrent event data with informative terminal event via   a gamma frailty by assigning a new type of double exponential Cox model to the terminal event. The main advantage of the proposed joint model is that it allows association between recurrent and terminal events and meanwhile, it overcomes the drawback of  the marginal effects of covariates die out over time, which exists commonly in gamma frailty Cox models. A sieve maximum likelihood approach  is carried out for parameter estimation, and the Bernstein polynomials are employed to approximate the non-decreasing cumulative baseline functions.  EM algorithm is utilized for optimization.  Asymptotic properties of the estimators are provided. Simulation studies are conducted to evaluate the finite sample behavior of the proposed estimators. A real data of chronic heart failure patients from the University of Virginia Health System is analyzed for illustration.