RE: Mixture model
Dear Dennis - Some approaches for you to consider:
1. Brute Force Model: I hope there is some prior information about the CL
for poor and extensive metabolizers. I am also assuming the difference between
PM and EM is meaningful. In that case, simply assume that this subject is a PM
and estimate a fixed effect for the same. The objective function and
uncertainty around parameter estimates should support this model over no CYP2D6
model.
2. Kind-of-Bayesian Model: Assuming you have the prior information on CL,
provide a prior for the prevalence and magnitude of difference in PMs with
perhaps a low uncertainty. This approach formally recognizes prior information,
but materially wouldn't be different from the BFM.
Either case, in my opinion, you would want to do justice to the 11 subjects who
you think are non-PMers with respect to point estimate and variance. The single
"PM" subject is unlikely to add conclusive information especially when a key
covariate (genotype) is missing.
[cid:[email protected]]
Joga Gobburu, PhD, FCP, MBA | Professor | School of Pharmacy | School of
Medicine
Executive Director |Center for Translational Medicine
N407, 20 N Pine, Baltimore, MD-21201
Office: (410) 706-5907 | E-mail:
[email protected]<mailto:[email protected]>
To receive latest Pharmacometrics news subscribe at
http://www.pharmacy.umaryland.edu/ctm/
To unsubscribe, send email to
[email protected]<mailto:[email protected]>
Quoted reply history
From: [email protected] [mailto:[email protected]] On
Behalf Of Fisher Dennis
Sent: Thursday, July 12, 2012 4:25 PM
To: [email protected]
Subject: [NMusers] Mixture model
Colleagues
I am analyzing data in which there are two distinct populations as a result of
CYP2D6 deficiency. In one dataset, there are 18 subjects with rich data; one
of these subjects is markedly different. In that the incidence of 2D6
deficiency is reported to be < 10%, one would expect only 1-2 deficient
subjects in this sample (consistent with the data here).
I was planning to use a mixture model as part of the analysis. However, with
only one subject in the deficient population, I am not sure if that is
appropriate.
Does anyone have any relevant experience or insight into this issue?
Dennis
Dennis Fisher MD
P < (The "P Less Than" Company)
Phone: 1-866-PLessThan (1-866-753-7784)
Fax: 1-866-PLessThan (1-866-753-7784)
http://www.PLessThan.com
<<inline: image001.png>>