From: nplock@zedat.fu-berlin.de Subject: [NMusers] ADAPT II with SAMPLE option Date: Tue, 18 Apr 2006 17:14:13 +0200 (CEST) Dear NONMEM users, I would like to ask you for your help concerning the software ADAPT II. I have conducted a clinical trial with 34 individuals and would now like to find out what were the informative sampling time points of the study. It seemed to me like this should be possible using the SAMPLE option in ADAPT II. Now to get familiar with the software I tried to use a simple two compartment model coded with clearance terms. I chose 1 model input, 1 input event time, for which I entered a time (0) and a rate of 1200 (I´m still confused where to tell the software how long the drug is applied for, because if I only put in a rate this does not really clarify how much of the drug is given, or am I missing a point here? Could anybody comment on that?). I then chose 1 model output equation with 8 observations at time points 0, 0.5, 1, 1.5, 2, 3, 6, 8, all to be optimized. I entered lower and upper time constraints (0,24) and then entered parameter values (CL=6.6, Vc=36.6, KA=1.91, CLd=18.5, Vp=33.1, all to be estimated). Initial conditions were all zero and not estimated. SDinter was 0.01 and SDslope 0.166 (am I right that these corresponds to the square root of EPS in NONMEM error model Y=IPRED+IPRED*EPS(1)+EPS(2)?) Then I chose D optimality with maximum of 10000 iterations and got the result: Near Singular Design Matrix. Can anybody tell me what my problem is? In addition, is there any literature available on how to use the SAMPLE option (because that is somehow neglected in the user´s guide)? I would really appreciate if anybody could clarify what I´m doing wrong. Thanks a lot for your help! Best regards Nele _______________________________________________________ From: bulitta@ibmp.osn.de Subject: Re: [NMusers] ADAPT II with SAMPLE option Date: Tue, 18 Apr 2006 18:32:19 +0200 I remember that those chapters of the Adapt II user manual helped me in the past (chapters and pages refer to Release 4 of the Adapt manual): Examples for D-optimal sampling: Chapter 4.7 Example pk7: SAMPLE ? D-Optimal Design, page 91 Chapter 6.5 Example 2compcl: SAMPLE ? D-Optimal Design, page 167 Specification of dosage regimens: Chapter 2.3 Preparing Data to be used with ADAPT, page 13 Chapter 3.2 Model Inputs, Bolus Inputs and Others, page 34 Best regards & many greetings Juergen _______________________________________________________ From: "Stephen Duffull" sduffull@pharmacy.uq.edu.au Subject: RE: [NMusers] ADAPT II with SAMPLE option Date: Wed, 19 Apr 2006 08:02:40 +1000 Hi Nele I have no experience at using ADAPT/ADAPT II. But I can confirm that your model with 8 sampling times does not produce a singular matrix. Indeed I tried this example in POPT (for Population OPTimal design http://www.uq.edu.au/pharmacy/sduffull/POPT.htm) and your original 8 sampling times was sufficient to provide good estimates for all your fixed and random effects parameters (I perturbed the zero time to 0.01; and assumed a lognormal variance for all of your BSV terms of 0.1). And this optimizes fine as well. I used 1200 as the dose. However you mention a [constant] rate of 1200 - which doesn't exist in your first order input model: >> (CL=6.6, Vc=36.6, KA=1.91, CLd=18.5, Vp=33.1, all to be estimated) Since you have an additive error component then if the dose is actually lower than 1200 units (e.g. 12 or 1.2 units) then the information matrix will quickly become poorly conditioned (although not singular for MATLAB). I hope this is of general interest. Regards Steve ==================================================================== Stephen Duffull School of Pharmacy, University of Queensland, Brisbane 4072, Australia Tel +61 7 3365 8808, Fax +61 7 3365 1688, Email: sduffull@pharmacy.uq.edu.au www http://www.uq.edu.au/pharmacy/index.html?page=31309 Design: http://www.uq.edu.au/pharmacy/sduffull/POPT.htm MCMC: http://www.uq.edu.au/pharmacy/sduffull/MCMC_eg.htm University Provider Number: 00025B ===================================================================== _______________________________________________________ From: "Tham Lai San, PHCV2" phcv2@nus.edu.sg Subject: RE: [NMusers] ADAPT II with SAMPLE option Date: Wed, 19 Apr 2006 08:06:41 +0800 Dear Nele, If you chose the option model input, then your drug is given like an infusion, and you will need to enter 2 event times. One for start of drug administration and one for stop whereby rate=0. This will tell ADAPT II the actual duration of infusion. If you consider SDinter and Sdslope as similar to assay error, then that will be similar to using the random unexplained variable from NONMEM. Regards, Lai San Tham _______________________________________________________ From: nplock@zedat.fu-berlin.de Subject: RE: [NMusers] ADAPT II with SAMPLE option Date: Wed, 19 Apr 2006 08:48:31 +0200 (CEST) Dear all who answered to my question, thank you very much for your input, it all helped a lot. I now used the proper coding for infusion and left out the estimation for KA, and it ran beautifully. I´ll get back to you when problems get more complicated :-) Thanks again Nele _______________________________________________________