5 replies to this topic

[Helmut Schütz]

HI, some questions (see attached). • The 'Number of Predicted Values' in the 'Engine Settings' is grayed out. Had to introduce artificial time points with 'Missing' concentrations in the raw data in order to check whether the lag-time are fitted correctly. I know that's 'by design' – but why? Only 'classical WNL' models show a smooth curve; I miss such an option in PHX-models as well. To obtain a smooth curve I had to set up a simulation and overlay observed data in an XY-plot. • There seems to be no way to specify dose units in ASCII models. If I use the same mass as in the mass part of the concentration I had to specify initial estimates of V_F which are 100times higher than the results of fitting the two phases separately. BTW, for this drug V_F of 3000L is reasonable, whereas 300000L is not. ;-) • Mapping initial estimates of a simulated ASCII model to the final estimates (Parameter=Parameter, Initial=Estimate) does not work: Execution Error/Failed to fit model. THX Simon for the video; didn't know this trick before. Still had to copypaste to an internal worksheet. • Did not succeed in setting up a multiple dose ASCII-code (I know, I know…) • Now for the tricky one: How to translate such a model (SD and MD) to a PHX-model? P.S.: Please don't feel rushed in any way; I'm just playing around. ;-) [file name=2PhExpl.phxproj size=953368]images/fbfiles/files/2PhExpl.phxproj[/file]

Attached Files

•  2PhExpl.phxproj   931.02KB   316 downloads

Edited by Simon Davis, 22 March 2017 - 01:37 PM.

[Helmut Schütz]

Follow-up...

Succeeded in setting up a PHX-model. Linking of Theta's to a sim works - after excluding 'sdev0' in the DW.

Interesting that results are quite different:

images/fbfiles/images/ID008.png

Edit: Forgot to use the same constraints in PHX like in the ASCII-model; now the fits are more similar:

images/fbfiles/images/ID008_2.png

Attached Thumbnails

• 
• 

[Simon Davis]

perhaps I misinterpreted your first question "To obtain a smooth curve I had to set up a simulation and overlay observed data in an XY-plot."

Actually it's not necessary in PHX models to set up a simulation you can simply request "add table" and request the level of details in the different areas you're interested in e.g.

seq(0,4,0.1),seq(4.5,48,0.5)

gives richer predicted curve in the first 4 hours. and then every 0.5 hours to 48. /extranet/media/kunena/attachments/legacy/images/tabstate.jpg

Note that this can be obtained in fitting or simulation modes and is more efficient than spreading a set number of points evenly over the time course.

I see you found some gaps in ASCII - are you happy now to get everything you used in ASCII In the PHX model or is there some other specific question. I am sure you see the advantages e.g. in simulation of using the newer Phoenix Modelling Language

Simon

Attached Thumbnails

• 

[Helmut Schütz]

Hi Simon!

seq(0,4,0.1),seq(4.5,48,0.5) gives richer predicted curve in the first 4 hours. and then every 0.5 hours to 48.
Note that this can be obtained in fitting or simulation modes and is more efficient than spreading a set number of points evenly over the time course.

Hey that's nice (though units are not carried over). Nevertheless I still have to set up a new XY-plot based on this table. Call me old-fashioned, but I’m still too used to the smooth curves (of predicted values) I get directly in ASCII- and classical WNL-models.

are you happy now to get everything you used in ASCII In the PHX model or is there some other specific question. I am sure you see the advantages e.g. in simulation of using the newer Phoenix Modelling Language

No THX – nothing at the moment ;-)

[Simon Davis]

I had a look at recording some enhancements from this discussion and I noted that QC 11734 requests "automatically getting smooth predicted curves" from Phx models - will need to see how exactly we can implement this request.

 

 Simon.

[Helmut Schütz]

Hi Simon!

Actually I like the method with seq(from,to,step) very much. It would be great if it’s possible to include an optional plot within the PHX-model.