Bayesian Inference of Interval-Censored Survival Data - Interval-Censored-Time-to-Event Data: Methods and Applications - page 182

the better the t. The second term 2p D in Equation (7.19) is the dimension
penalty. The DIC in Equation (7.19) is a Bayesian measure of predictive model
performance, which is decomposed into a measure of t (D( )) and a mea-

sure of model complexity (p D ). The smaller the value, the better the model
will predict new observations generated in the same way as the data. Other

properties of the DIC can be found in Spiegelhalter et al. (2002).

7.5 \Dynsurv" Package

The dynamic model of Wang et al. (2011a) and the time-varying coecients

Cox model of Sinha et al. (1999), along with some variants, are implemented in

C++ based on the Boost C++ library (e.g., Karlsson, 2005). A user-friendly

interface to R (R Development Core Team, 2011) is provided via the R package

dynsurv(Wang et al., 2011b), which is publicly available at http://cran.
r-project.org/web/packages/dynsurv/ .
The main function that fits the Bayesian models to an interval-censored

data set is bayesCox . Its usage is

bayesCox(formula,data,grid,out,
model=c( `` TimeIndep", `` TimeVarying", `` Dynamic"),

base.prior=list(),coef.prior=list(),
gibbs=list(),control=list())

The arguments of the function are briefly explained as follows:

•

formula : a formula object. The response must be a survival object as
returned by the Surv function from thesurvivalpackage.

• data : a data.frame containing variables used in formula .

• grid : vector of prespecified time grid points.