
162 Current Trends in Bayesian Methodology with Applications
0
(a)
time
log viral load
0 50 100 150 200
2 3
4 6
(b)
time
log viral load
Mean of group
Patient 12
Patient 13
Patient 22
Patient 29
Patient 36
Patient 42
50 100 150 200 0 10 20 30 40
2
3 4 5
6
5
0 5 10 15 20 25 30
(c)
Index
K−L divergence
FIGURE 8.1
ACTG 315 data. (a) Viral lo ads in log
10
scale. (b) Individual profiles of some
influential observations under the normal SME model. (c) KL distance under
the nor mal SME model.
is achieved by using the cubic spline basis to represent f(t
ij
) as α
0
+ α
1
t
ij
+
α
2
t
2
ij
+ α
3
t
3
ij
+
P
K
s=1
c
s
|t
ij
−κ
s
|
3
+
. Thus, the non-parametric component N
i
f
i
from the model defined