Prediction of Time to a Terminal Event (TTTE) of New Units in a Dynamic Recurrent Competing Risks Model
share
In this paper, we propose a simulation approach to predict time to terminal event (TE) that arises from joint dynamic modelling. Many joint dynamic models have found applications in medical research. Predicting terminal event times is often in the most interest due to its value as a prognostic tool in medical treatments and the complexity in developing appropriate prediction methodology. When a joint dynamic model gets more complicated, the computational aspect of predicting TE can be particularly challenging. An alternative, which is simulating censored event times according to history of data accrual has not been considered in previous works, to our knowledge. Based on the class of joint dynamic models of recurrent competing risks (RCR) and terminal event (TE) in Liu and Peña (2015), we demonstrate how to predict terminal event (TE) times by the simulation approach. We also point out the size-biased sampling related to gap time that traverses monitoring time.
READ FULL TEXT