A1265
Title: Beyond hazard-based illness-death models for semi-competing risks data
Authors: Liming Xiang - Nanyang Technological University (Singapore) [presenting]
Rui Huang - Nanyang Technological University (Singapore)
Abstract: The illness-death model has been developed as a common approach for regression analysis of semicompeting risks data by modeling transitions between states using the proportional hazards or accelerated failure time model. An illness-death model is proposed that is formulated based on the mean residual life function, offering a more straightforward and interpretable framework for modeling state transitions in the semicompeting risks setting. Estimation is facilitated through novel estimating equations derived from a penalized quasi-likelihood approach incorporating inverse probability weighting. The method requires no distributional assumption on the latent frailty term, thereby reducing the risk of model misspecification. Simulation studies demonstrate its promising performance across a range of realistic scenarios and an application to a myeloma progression study illustrates its practical utility.
