NONMEM Users Network Archive

Hosted by Cognigen

RE: Inflated random effects showed by VPC

From: Jiang, Yu <yu-jiang-1>
Date: Thu, 4 Sep 2014 17:30:43 +0000

Dear Ken,

Thank you for your suggestion! That's a good point. But I have tried the om=
ega block before and it didn't help with the VPC plot. I am currently using=
 Michalis-Menten kinetics and parameterize it as Vm*(A(2)/V)/(Km+(A(2)/V)).=
 I didn't put inter-individual variability on Km(constrained by data) so on=
ly left etas of V and Km. Those two etas don't seem to correlate with each =
other from the scatter plots. I tried incorporate the covariance term and t=
he covariance term is estimated to be near zero. So I think there should be=
 other problems.

From: Ken Kowalski [ken.kowalski
Sent: Thursday, September 04, 2014 11:27 AM
To: Jiang, Yu; nmusers
Subject: RE: [NMusers] Inflated random effects showed by VPC

Dear Yu,

Did you explore block Omega structures to investigate potential correlation=
s among the random effects? If you assumed a diagonal Omega structure wher=
e the random effects are assumed to be independent when they are indeed cor=
related this can inflate the between-subject variability in your simulation=
s of the concentrations. For example, if the IIV random effects for CL and=
 V are highly correlated but you simulate assuming these random effects are=
 independent then you will likely simulate some extreme combinations of sub=
ject-specific values of CL and V that may not be represented in your data.


Kenneth G. Kowalski
President & CEO
A2PG - Ann Arbor Pharmacometrics Group, Inc.
110 Miller Ave., Garden Suite
Ann Arbor, MI 48104
Work: 734-274-8255
Cell: 248-207-5082
Fax: 734-913-0230

From: owner-nmusers
 Behalf Of Jiang, Yu
Sent: Thursday, September 04, 2014 12:15 PM
To: nmusers
Subject: [NMusers] Inflated random effects showed by VPC

Dear all,

I wonder what might cause a pharmacokinetic model to have inflated variabil=
ity. For my model, the GOF plots look reasonably good--meaning that the fix=
ed effects are OK. However the prediction corrected VPC implemented by PsN =
indicated severely overestimated variability regardless of whether I strati=
fy them into different dose groups or analysis them altogether. I have trie=
d all my candidate models, all of them have the observed 95th and 5th perce=
ntile way off from the simulated confidence bands, and for some of them, th=
e observation points don't even go into the upper and lower confidence inte=
rval bands.

I checked my eta plots, and I think although they don't look perfectly norm=
al, they still looks reasonably symmetrical with a bell shape. Eta on V see=
ms to be a little skewed to the right. I don't have much experience on PopP=
K so I might be wrong.

I think there might three possibilities causing this problem.

One is that, the true distribution of etas is not normally distributed but =
more like uniformly distributed (or skewed). The estimation step have no pr=
oblem of identifying the right mean and variance for parameters even the tr=
ue underlying distribution is not normal distribution. But when it comes to=
 simulation, the simulated parameters are draw from the normal distribution=
 with the estimated mean and variance. That discrepancy might cause inflate=
d variability in simulated parameters and therefore inflated variability in=
 simulated observations.

The other is that there are a few subjects having very large eta compared w=
ith other subjects, therefore inflated the estimated omega.

Also all my subjects are dosed based on their weight, height, gender and ag=
e to achieve a target drug concentration level. They might do a very good j=
ob making the concentrations to reach the target level so all of my observa=
tions lies in the middle of the prediction corrected VPC plots. I think thi=
s is the least likely possibility since I have already taken covariate effe=
cts into consideration in some of my models..

I am not sure I am thinking it right. Please correct me if I am wrong. Does=
 anyone have any thoughts into this? Has anyone encountered similar things =
before? I truly appreciate any comments or suggestions.


Graduate student in Clinical Pharmaceutical Science

University of Iowa

Received on Thu Sep 04 2014 - 13:30:43 EDT

The NONMEM Users Network is maintained by ICON plc. Requests to subscribe to the network should be sent to:

Once subscribed, you may contribute to the discussion by emailing: