| |
 |
|
|
Science Forum Index » Statistics - Math Forum » Arjas plot...
Page 1 of 1
|
| Author |
Message |
| Tobias Janstad... |
 Posted: Sun May 04, 2008 11:02 am |
|
|
|
Guest
|
#I got some problems with Arjas plot. Try to reproduce #figure in textbook.
library(splines);
library(survival);
library(KMsurv);
data(bmt)
#143 bone marrow patients
#goal: reproduce figure 11.15 in Klein Moeschberger survival analysis (1997).
#(i) divide the data material into two subgroups
MTX<-subset(bmt,z8==1)#MTX used as an GVHD prophylactit.
noMTX<-subset(bmt,z8==0)#
#(ii) sort with respect to survival time within each group
#allready done
#(iii)
#N_g(t_i), following notation from page 370 in K.M. 1997
MTXcumdelta<-cumsum(MTX$d3)
noMTXcumdelta<-cumsum(noMTX$d3)
coxMTX=coxph(Surv(MTX$t1,MTX$d3)~z1+z2+z3+z4+z5+z6+z7,data=MTX,method="breslow")
print("Cox-model fitted for MTX-grupp.")
coxnoMTX=coxph(Surv(noMTX$t1,noMTX$d3)~z1+z2+z3+z4+z5+z6+z7,data=noMTX,method="breslow")
print("Cox-model fitted for noMTX-grupp.")
MTXbase<-basehaz(coxMTX,centered=TRUE)#estimate the cumulative baseline hazard (I guess)
noMTXbase<-basehaz(coxnoMTX,centered=TRUE)
MTXcumbase<-cumsum(MTXbase$hazard)
noMTXcumbase<-cumsum(noMTXbase$hazard)
plot(stepfun(MTXcumdelta[1:length(MTXcumbase)],c(0,MTXcumbase)))
lines(stepfun(noMTXcumdelta[1:length(noMTXcumbase)],c(0,noMTXcumbase)))
lines(c(0,20),c(0,20))
#Why do length of cumulative hazard and events differ? How can this be?
#The plot does not look like figure 11.15. Can someone help me correct it? |
|
|
| Back to top |
|
| |
|
Page 1 of 1
All times are GMT - 5 Hours
The time now is Sun Sep 07, 2008 12:19 pm
|
|