Population PKPD Mode.. - Pharsight Corporation

What to use in diagnostics ?
●
⎡Ω11
Ω=
⎢
⎢
Ω Ω
⎢⎣
0 0
Simulation Based Diagnostics
• Posterior Predictive Checks
1
Θ=
⎢
θ
⎥
⎢
2 ⎥
⎢θ
⎥
3
12 22
⎡σ
♦
♦
♦
♦
♦
Ω
33
⎤
⎥
⎥
⎥⎦
⎤
11
Σ= ⎢
0 σ ⎥
22
⎣
⎡θ
⎤
⎣
⎦
⎦
Select a statistic of interest that a good model should not miss (e.g. AUC,
Cmax, etc) and that can be derived from the raw data
Simulate many data sets from the model (with uncertainty, TS2.2)
Calculate the statistic from the N simulated data and J Replicates
Calculate the statistic from the observed data
Compare the statistic from raw and simulated data
⎡Var
⎤
⎢
Cov Var
⎥
VARCOV = ⎢
⎥
⎢ Cov Cov Var ⎥
⎢
⎥
⎣Cov Cov Cov Var⎦
Draw N Parameters form the
final estimates and Variance
Covariance Matrix
Simulate Data according
to the design of the study
and compute the statistic
Compare the statistic
from raw and simulated
data
If we ignore parameter uncertainty this reduces to predictive check
slide 47
© Pharsight Corporation All Rights Reserved