70 Exposure-Response Modeling: Methods and Practical Implementation
4.2.3 The argatroban example
We use the argatroban da taset to illustrate applications of joint modeling
of longitudinal data. SAS proc NLMIXED is used for the analysis but the
approach of implementation can be applied to similar packages such as NON-
MEM and R. Detailed example codes can be found in the appendices. In
this exa mple, both exposure and response are continuous variables and can
be assumed to have a normal distribution. Therefore, we can use the mean
and variance in the model s w itching between the exposure and response. The
following codes run the joint mo de l:
c
∗
ij
=
d
i
(1 − exp(−t
ij
Cl
i
/V
i
)) t
ij
< t
f
d
i
(1 − exp(−t
f
Cl
i
/V
i
)) exp(−(t
ij
− t
f
)Cl
i
/V
i
) t
ij
≥ t
f
c
ij
= c
ij
+ e
ij
y
ik ...