NONMEM Users Network Archive

Hosted by Cognigen

Re: var-cov matrix issue?

From: Ethan Wu <ethan.wu75>
Date: Tue, 24 Feb 2009 11:09:08 -0800 (PST)

Hi Justin, only ETA was estimated with high SE but, again, I guess it cam=
e back to the question: how trustful it is if such error message appears=

artis.com" <justin.wilkins
 Tuesday, February 24, 2009 1:19:17 PM Subject: Fw: [NMusers] var-cov mat=
rix issue? Dear Ethan, Algorithmically singular matrices are =
often a sign that that your model is ill-conditioned in some way; I would b=
e careful in how I used the variance-covariance matrix in this scenario, an=
d especially for simulation. Are there any parameters that are being estima=
ted with particularly high standard errors? This might suggest overparamate=
rization. Not sure how helpful this is! Best Justin Jus=
tin Wilkins Senior Modeler Modeling & Simulation (Pharmacology) CHBS,=
 WSJ-027.6.076 Novartis Pharma AG Lichtstrasse 35 CH-4056 Basel Swi=
tzerland Phone: +41 61 324 6549 Fax: +41 61 324 3039 Cell: +41 76 561=
 0949 Email : justin.wilkins
 Wilkins/PH/Novartis on 2009/02/24 07:15 PM ----- Ethan Wu <ethan.wu7=
5
M To nmusers
sue? Dear all,  I recently encounter this error mess=
age (below). My objective was to use nonmem var-cov output  for approxima=
tion of distribution of parameters for performing a simulation.  if su=
ch error message occur, is the var-cov matrix  still OK to use? -- I k=
now that better way to figure out distribution of parameters is to do boots=
trap, but given limited time I have....   thanks   "0MINIMI=
ZATION SUCCESSFUL NO. OF FUNCTION EVALUATIONS USED:  331 NO. OF SIG. =
DIGITS IN FINAL EST.:  3.3  ETABAR IS THE ARITHMETIC MEAN OF THE ETA=
-ESTIMATES, AND THE P-VALUE IS GIVEN FOR THE NULL HYPOTHESIS THAT THE TRU=
E MEAN IS 0.  ETABAR:   0.11E-02 SE:       0.23E-01  P =
VAL.:   0.96E+00 0S MATRIX ALGORITHMICALLY SINGULAR 0S MATRIX IS OUTP=
UT 0INVERSE COVARIANCE MATRIX SET TO RS*R, WHERE S* IS A PSEUDO INVERSE O=
F S 1 "
Received on Tue Feb 24 2009 - 14:09:08 EST

The NONMEM Users Network is maintained by ICON plc. Requests to subscribe to the network should be sent to: nmusers-request@iconplc.com.

Once subscribed, you may contribute to the discussion by emailing: nmusers@globomaxnm.com.