Biomedical Engineering Reference
In-Depth Information
The argument
data
corresponds to an input data set. It consists of five
columns whose names are
id
,
time
,
tumor
,
death
, and
trt
. The column
id
in
data
denotes an identification number,
time
a death time or a sacrifice
time,
tumor
a 0-1 indicator whether or not a tumor onset occurs (0: without
a tumor onset, 1: with a tumor onset),
death
a 0-1 indicator whether or
not a death occurs (0: sacrificed, 1: died), and
trt
a binary covariate. The
argument
int
is breaking time points for piece and does not include time
points 0 and the maximum time which is set in advance. For example, when
the maximum observed value is 12,
int=c(3,6,9)
implies that the interval
(0,12] is split into four sub-intervals such as (0,3], (3,6], (6,9], (9,12]. In the
argument
frailty
, we can choose `
gamma
' to t a gamma frailty model or
`
normal
' to t a normal frailty model. Here, the gamma and normal frailties
follow G(
1
;) and N(0;); respectively. The argument
init
denotes the
initial value of each parameter (default = 0 for
1
;
2
; and
3
; 0.1 for
01
(t)
and
02
(t); and 1 for ); and the argument
tol
the convergence criterion of
EM algorithm (default = 0.001). Finally, the argument
plots
requires that
we plot the estimates of
1
;
2
;
3
; and through the EM algorithm until the
convergence criterion is satisfied (default = F).
For a chosen frailty, the packageCSD, by default, produces the estimates
for the regression parameters,
1
;
2
; and
3
; the baseline hazard rates for
tumor onset time and death time,
01
(t) and
02
(t); and (denoted as
Estimate
) along with their standard errors (denoted as
SE
), the 95% con-
fidence limits (denoted as
(L,U)
), test statistic (denoted as
Z
), and the result-
ing p-values (denoted as
P-value
). Finally, it provides the values of the full
log-likelihood and sample size used in the analysis.
Search WWH ::
Custom Search