A0178
Title: Flexible accelerated time modeling of recurrent events data in the presence of a dependent terminal event
Authors: Limin Peng - Emory University (United States) [presenting]
Abstract: Accelerated time modeling provides a useful prospective for assessing covariate events on recurrent event outcomes that have physical interpretations. The generalized accelerated recurrence time model (GART) significantly extends the traditional accelerated failure time model for recurrent events, offering extra flexibility in accommodating heterogeneous covariate effects. In practice, the observation of recurrent events is often stopped by a dependent terminal event. To address such a realistic scenario, we discuss two extensions of the GART models that can appropriately account for the presence of a dependent terminal event. We develop estimation and inference procedures for both extensions of the GART model, and establish desirable asymptotic properties. The proposed estimation and inference procedures can be readily implemented based on existing software. Simulation studies demonstrate satisfactory finite-sample performance of the proposed methods. We illustrate the proposed methods via an application to a dataset from the Cystic Fibrosis Foundation Patient Registry (CFFPR).
