gofM.phreg: GOF for Cox covariates in PH regression

In mets: Analysis of Multivariate Event Times

GOF for Cox covariates in PH regression

Description

Cumulative residuals after model matrix for Cox PH regression p-values based on Lin, Wei, Ying resampling.

Usage

gofM.phreg(
  formula,
  data,
  offset = NULL,
  weights = NULL,
  modelmatrix = NULL,
  n.sim = 1000,
  silent = 1,
  ...
)

Arguments

formula	formula for cox regression
data	data for model
offset	offset
weights	weights
modelmatrix	matrix for cumulating residuals
n.sim	number of simulations for score processes
silent	to keep it absolutely silent, otherwise timing estimate will be prduced for longer jobs.
...	Additional arguments to lower level funtions

Details

That is, computes

U(t) = \int_0^t M^t d \hat M

and resamples its asymptotic distribution.

This will show if the residuals are consistent with the model. Typically, M will be a design matrix for the continous covariates that gives for example the quartiles, and then the plot will show if for the different quartiles of the covariate the risk prediction is consistent over time (time x covariate interaction).

Author(s)

Thomas Scheike and Klaus K. Holst

Examples

library(mets)
data(TRACE)
set.seed(1)
TRACEsam <- blocksample(TRACE,idvar="id",replace=FALSE,100)

dcut(TRACEsam)  <- ~. 
mm <- model.matrix(~-1+factor(wmicat.4),data=TRACEsam)
m1 <- gofM.phreg(Surv(time,status==9)~vf+chf+wmi,data=TRACEsam,modelmatrix=mm)
summary(m1)
if (interactive()) {
par(mfrow=c(2,2))
plot(m1)
}

m1 <- gofM.phreg(Surv(time,status==9)~strata(vf)+chf+wmi,data=TRACEsam,modelmatrix=mm) 
summary(m1)

## cumulative sums in covariates, via design matrix mm 
mm <- cumContr(TRACEsam$wmi,breaks=10,equi=TRUE)
m1 <- gofM.phreg(Surv(time,status==9)~strata(vf)+chf+wmi,data=TRACEsam,
		  modelmatrix=mm,silent=0)
summary(m1)

mets documentation built on May 29, 2024, 3:51 a.m.