B1918
Title: Multiple longitudinal joint model with informative time measurements
Authors: Ines Sousa - Minho University (Portugal) [presenting]
Abstract: In classical longitudinal models the longitudinal observed process is considered independent of the times when measurements are taken. However, in medical context it is common that patients in worst health condition are more often observed, whereas patients under control do not need to be seen so many times. Therefore, longitudinal models for data with this characteristic should allow for an association between longitudinal and time measurements processes. In this work we consider a response longitudinal variable with Gaussian distribution. We propose a model where the follow-up time process is stochastic. The model is described through the joint distribution of the observed process and the follow-up time process. Estimation of model parameters is through maximum likelihood. We conducted a simulation study of longitudinal data where model parameter estimates are compared, when using the model proposed and ignoring the association between processes. Finally, the model proposed is applied to a real data set when monitoring for biomarkers CEA and CA15.3 on breast cancer progression. In this case the follow-up time process should be considered dependent on the longitudinal outcome process. Results are presented showing that, ignoring the latent process of time measurements brings bias results when the collected time points are associated with the observed process.
