Re: [NMusers] Correlation Test in NONMEM

From: Jian Xu <alanub_at_yahoo.com>
Date: Thu, 16 Aug 2007 14:58:44 -0700 (PDT)

Hi, Jeroen, Thanks for your reply. Are you talking about the shrinkage=
, which is recently published by Dr. Karlsson's group? I will check it out=
, and too much shrinkage may indicate my model is overparameterized even I =
did not see any estimate is close to zero. I did plot just two etas toget=
her. Before covariance added, it was positively correlated, but after run, =
it became negatively correlated. So weird. Jian ----- Original Me=
ssage ---- From: "Elassaiss - Schaap, J. (Jeroen)" <jeroen.elassaiss_at_orga=
non.com> To: Jian Xu <alanub_at_yahoo.com> Sent: Thursday, August 16, 2007=
 5:47:13 PM Subject: RE: [NMusers] Correlation Test in NONMEM =
  DIV { MARGIN:0px;} Dear Jian, Have you checked=
 for shrinkage of your etas? That could be a source of falsely inducemen=
t of correlation. A way to check your plot (you say delta plots) is to=
 simply plot the two etas against other. Best regards, =
Jeroen From: owner-nmusers_at_globomaxnm.com [mailto:owner-nm=
users_at_globomaxnm.com] On Behalf Of Jian Xu Sent: Thursday, 16 August, =
2007 21:28 To: nmusers_at_globomaxnm.com Subject: [NMusers] Correlation=
 Test in NONMEM Dear NM users, During a pop PK =
model development, I found two parameters (based delta plots) are highly=
 correlated (positive correlation), therefore, I tried to add Covariance=
 between them in order to account for this correlation. However, after t=
his run, it turns out to be a negative correlation. Anybody had this exp=
erience before? Could it be due to the model misspecification? Or datase=
t does not support the IIV on these two parameters? Any input is appre=
ciated. Thank you in advance. Jian Shape Yahoo! in=
 your own image. Join our Network Research Panel today! This me=
ssage, including attachments, is confidential and may be privileged. If you=
 are not an intended recipient, please notify the sender then delete and de=
stroy the original message and all copies. You should not copy, forward and=
/or disclose this message, in whole or in part, without permission of the s=
ender. ______________________________________=
______________________________________________ Be a better Globetrotter. =
Get better travel answers from someone who knows. Yahoo! Answers - Check it=
 out. http://answers.yahoo.com/dir/?link=list&sid=396545469
Received on Thu Aug 16 2007 - 17:58:44 EDT

This archive was generated by hypermail 2.2.0 : Tue Nov 06 2007 - 15:06:58 EST