Hi All, I’m having trouble getting NLME to calculate standard error for the mixed residual error model. Here is part of the Core Output that I am getting. The only paramater value need a limit would be stdev0 and I know of no way to change that limit. Is this a problem whenever I use the mixed or combined residual error model? Kind regards, Jim Starting std error computation Warning - following parameter value near bound No standard error computation possible i Bound low Parameter (i) Bound high 6 0.1000E-02 0.1000E-02 0.1000+101 Std errors aborted One or more parameter values at a bound Engine Convergence Summary Convergence achieved OPTIF9 ITRMCD value on last iteration = 1 NLME7 return code= 1 Optimal parameters: 0.30735426E+05 1 1 THETA BOUNDS: 0.000E+00 0.100+101 0.32796467E+05 2 1 THETA BOUNDS: 0.000E+00 0.100+101 0.12975871E+04 3 1 THETA BOUNDS: 0.000E+00 0.100+101 0.10251705E+04 4 1 THETA BOUNDS: 0.000E+00 0.100+101 0.22310071E+03 5 1 THETA BOUNDS: -0.100+101 0.100+101 Warning - following parameter value near bound No standard error computation possible i Bound low Parameter (i) Bound high 6 0.1000E-02 0.1000E-02 0.1000+101
Hi Jim, I believe what you’re referring to as a ‘standard error’ is actually the estimate of the std deviation associated with eps in the residual error model. Getting such a small value suggests you’re getting nearly a perfect fit - that is, the model predictions lie riught on top of the observations - is that what you’re observing? (There is a built in lower limit of .001 on the eps value). If you could post (or send Support) the model/data, we’d be happy to take a look at it. Simon.
Hi Simon,
Yes, that is correct. It is the residual error estimation that seems to be causing the problem, but it is reported under the standard error calculations section of the output.
As this is client data you are asking me to share, I’m understandablly hesitant. However, I could share output with you if you wish. This is a study design where a initial dose is given by IV infusion followed by a number of optional top-up doses. The entire study has samples over about 4 half-life values but with most of the samples during the first 2 half-life values.
Hiere is a figure of a poorly fit set of data. Let me know how I can share data with support without posting it on the forum. Let me know what output files you want to see.
[img/] [file name=Drug_Figure_for_Pharsight.docx size=55849]Certara | Drug Development Solutions