|
Abstract:
|
In this paper, I will consider extensions of the standard joint model for longitudinal and survival data to more complex situations. This work is motivated by applications in radiation therapy of prostate cancer, in which longitudinal PSA and clinical endpoint data are collected. The extensions of the standard model include incorporation of a cured fraction, use of CPO statistic to assess the need for a cured fraction, nonlinear longitudinal models, comparison of non-nested models, hierarchical models for data from multiple institutions, covariate selection, time-dependent hazard models that involve more than just the current value of the longitudinal variable, and use of the models for individualized predictions.
|
