A1200
Title: Bayesian joint modelling of longitudinal and competing risks data with cause dependent masking
Authors: Mahaveer Singh Panwar - Banaras Hindu University, Varanasi (India) [presenting]
Abstract: Joint modelling of longitudinal measurements and time-to-event data has received considerable attention, especially in the field of public health studies. The longitudinal and competing risks event processes, with a linear mixed effect model and cause-specific hazard model, are jointly analyzed, respectively. The two processes are associated with the shared random effect approach. In the competing risks data, the event's cause is not always known, leading to incomplete data concerning the cause. The cause of events for such individuals is said to be masked. It is assumed that the masking is not independent of the causes, and hence, our proposed joint model also deals with cause-dependent masking situations in competing risk data. The estimation of model parameters is carried out under the Bayesian paradigm as it is computationally more flexible against the model complexity. An extensive numerical study is performed to evaluate the efficacy of estimators obtained under the joint model. The Prostate, Lung, Colorectal, and Ovarian (PLCO) cancer prevention trial dataset illustrates the established methodology subsuming the dependent masking in competing risks data.
