A0892
Title: Regression and prediction for competing risks with doubly truncated data
Authors: Jacobo de Una-Alvarez - University of Vigo (Spain)
Carla Moreira - University of Minho (Portugal) [presenting]
Abstract: Regression and prediction for competing risks have been traditionally performed through a proportional hazards assumption. Such proportionality assumption can be established for the cause-specific hazards or transition intensities. In this objective, the approach is investigated when the target time is doubly truncated. More specifically, the applicability of the inverse probability weighting approach presented recently in literature in the competing risks setup is studied. The properties of the introduced procedures are investigated through simulations. Estimation, testing and prediction issues are considered. For illustration purposes, the proposed methods are applied to real data.
