NONMEM Users Network Archive

Hosted by Cognigen

AAPS free webinar: The Full Covariate Models and WAM Algorithm

From: Zhang, Liping <liping.zhang3>
Date: Tue, 6 Apr 2010 10:05:17 -0400

Sponsored by the Population Pharmacokinetics and Pharmacodynamics Focus Gro=

American Association of Pharmaceutical Scientists


* Wednesday, Apr 14, 2010 from 12:30 - 2:00 pm EST



* The Full Covariate Models and WAM Algorithm:

*Efficient Building of Covariate Models and Appropriate Inferences about Co=
variate Effects



* Conducted by

* Marc R. Gastonguay, Ph.D., Metrum Research Group & Metrum Insti=

* Kenneth G. Kowalski, M.S., Ann Arbor Pharmacometrics Group

* Moderated by

* Liping Zhang, Ph.D., Bristol-Myers Squibb



log in 15 min before it starts, at

The Q&A Session will follow the formal lecture at approximately 1:15 pm EST=
. You may ask questions at any time during the webinar by typing them in to=
 the question box on your screen. You may also ask questions in advance dur=
ing the registration process.

PDF Handouts: you may download the presenter's handouts at:

Firewall or Other Connection Issues? Please visit the following site for tr=
oubleshooting information:

Past Problems Logging In?

Here is a URL which will help you connect to our webinars: www.gotomeeting.=


This presentation will provide an overview of Full Covariate Modeling metho=
d (Gastonguay, 2004) and Wald's Approximation Method (WAM) for covariate mo=
del building based on the methodology of Kowalski and Hutmacher (2001). An =
example will be used to illustrate application of the WAM algorithm and the=
 value obtained from evaluating all reduced models among the complete set o=
f possible hierarchical covariate models leading to a parsimonious final mo=
del. The presentation will also illustrate the utility of the Full Covariat=
e Model approach for inference about covariate effects and some possible st=
rategies for the successful development of a full model necessary to employ=
 the WAM approach and subsequent model-based simulation goals. The presenta=
tion will highlight the advantages and disadvantages of the Full Covariate =
Model approach and WAM approach and in comparison to standard stepwise proc=
edures. The presentation will conclude with a discussion of the available s=
oftware to implement the WAM algorithm and Full Covariate Model.

The WAM algorithm was first published in 2001; however, it has not been wid=
ely used in the modeling community in part because it requires the developm=
ent of a full model. At the time of its introduction, development of a full=
 model in which all covariate effects are estimated simultaneously was not =
routinely performed. Today, there is a greater appreciation of the value in=
 developing full models, particularly for inferential purposes regarding co=
variate effects. This webinar will present two promising approaches for cov=
ariate model building that leverage information from a full model as altern=
atives to standard stepwise procedures.

Goals and Objectives:

* To provide an overview of the goals of covariate modeling in the drug =
development process;

* To provide an overview of Full Covariate Model and WAM algorithm appro=
aches relative to traditional Stepwise Regression, define advantages and di=
sadvantages of each method as a function of covariate modeling objectives i=
n drug development;

* To promote an awareness of the methodologies and the available softwar=

This message (including any attachments) may contain confidential, propriet=
ary, privileged and/or private information. The information is intended to =
be for the use of the individual or entity designated above. If you are not=
 the intended recipient of this message, please notify the sender immediate=
ly, and delete the message and any attachments. Any disclosure, reproductio=
n, distribution or other use of this message or any attachments by an indiv=
idual or entity other than the intended recipient is prohibited.

Received on Tue Apr 06 2010 - 10:05:17 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: