Biomedical Engineering Reference
In-Depth Information
5.5
Data Analysis
In this section, we apply the method proposed in the previous section to the
subset of data from ED
01
, which was conducted by the National Center for
Toxicological Research and involved 24,000 female mice. Following Lindsey
and Ryan (1994), we analyzed the lung tumor data set of 671 mice and two
treatments,; the control group and high-dose group of the ED
01
have 387
and 284 mice, respectively. First, read the data set stored in text file format
externally and save it into the object
lung
with the following command:
`
'
'
'
'
'
>lung=read.table(
lung.txt
,col.names=c(
id
,
time
,
'
'
'
'
tumor
,eath
,rt
))
Load a source file
CSD.r
containing the package
CSD
and execute the package
with
lung
and
int=c(12,18)
for each frailty model as follows:
>CSD.gamma=CSD(data=lung,int=c(12,18),frailty=
'
gamma
'
,
init=c(0,1,0.1),tol=0.001,plots=F)
>CSD.normal=CSD(data=lung,int=c(12,18),frailty=
'
normal
'
,
init=c(0,1,0.1),tol=0.001,plots=F)
The contents of
CSD.gamma
and
CSD.normnal
are displayed in Tables 5.1
and 5.2, respectively.
Because the results are similar to each other in terms
of estimates and their standard errors,we only focus on those for the gamma
frailty model. The estimates of
01
(t) are 0.009 (SE = 0.001), 0.005 (SE =
0.001), or 0.053 (SE = 0.010) and those of
02
(t) are 0.002 (SE = 0.001),
0.007 (SE = 0.002), or 0.083 (SE = 0.011) for t 2(0,12], (12,18], or (18,33],
respectively. The regression parameters are estimated as
1
= 0:202;
2
=
0:157; and
3
= 0:991 with the p-values of 0.275, 0.454, and less than 0.001,
respectively. They imply that the effects of treatment on both tumor onset
Search WWH ::
Custom Search