Bayesian Joint Modeling of Patient-Reported Outcomes and Survival Information: Discussion
A. Lawrence Gould
Joint modeling of longitudinal patient-reported outcomes and corresponding survival endpoints is becoming ever more crucial in practice. Full utilization of this data enables better understanding of the underlying relationships between ongoing measurements and clinical outcome, as well as obtain more efficient, potentially adaptive trial designs. This session will contrast Bayesian and frequentist methods for joint modeling of such datasets, with particular emphasis on how much extra learning about parameters of interest one can gain by going to the trouble of joint modeling. The expertise in the session will be drawn from a new Bayesian scientific working group intended to build bridges between the Drug Information Association (DIA) and the statistical community, and will feature speakers from both academics and industry.
