A1653
Title: On the robustness of maximum h-likelihood estimators under cause-specific competing risks frailty models
Authors: Il Do Ha - Pukyong National University (Korea, South) [presenting]
Hangbin Lee - Chungnam National University (Korea, South)
Youngjo Lee - Seoul National University (Korea, South)
Chew Chee - Universiti Malaysia Terengganu (Malaysia)
Abstract: In clustered competing risks data, both the events of interest and competing events are often correlated within the same cluster. This dependency can be modeled using cluster-specific frailties and the parameters can be estimated by maximum h-likelihood estimators (MHLEs). The robustness of MHLEs under the common log-normal frailty assumption is theoretically investigated when the true frailty distribution deviates from log-normality. A robust non-parametric method for cause-specific competing risks frailty models is proposed, which does not assume a specific parametric form for the frailty distribution. The proposal includes a hybrid iterative algorithm for computing non-parametric maximum likelihood estimators (NPMLEs). Simulation studies under various frailty distributions indicate that the proposed NPMLEs demonstrate superior performance, particularly for the association parameter. MHLEs of the regression coefficients yield comparable performance in terms of average relative bias and root mean squared error, which aligns with theoretical findings. Both methods are applied to real-world datasets from multi-center studies on bladder cancer and bone marrow transplantation to demonstrate their practical utility.
