New FDE demo - Continuous inputs and outputs

Question

Hello everyone,

Peter Hersh and I are developing an FDE demo with both continuous inputs and outputs.  The premise is that glucose levels were monitored in patients before-and-after a treatment was administered.  There are five treatments in total.  The objective is to predict the glucose response of a new patient to a treatment based on their baseline glucose level.  We recorded an initial 10-minute version of the demo here.  The PowerPoint, JMP Projects archive and patient files are included below.  Please take a look and let us know if you have any feedback.

P.S. turns out JMP Projects Archive files are not supported as file attachments in the Community.  I've therefore include the Project file itself, and all referenced files.  I don't know how easy it will be to put it all together.  Let me know if you want the Archive file and I'll email it to you.

Patient_9997.jmp

Patient_9999.jmp

Patient_9998.jmp

After - B-Spline Model Summaries (Degree=3, Knots=20).jmp

Glucose_before.jmp

Glucose_after.jmp

Load Patient Data.jrn

Glucose Project.jmpprj

Functional Data Analysis Glucose .pptx

HadleyMyers · Answer

Hi @stan_koprowski

Thank you, I tried just now but for some reason it still doesn't seem to work for me (see error message below).

stan_koprowski · Answer

Hi @HadleyMyers ,

Sorry you had some difficulties with trying to attach a JMP project archive to your post.

There was a bug in the community that prevented that type of attachment from being added to posts.

This has been fixed.

Feel free to try agin if you would like to edit the post to include only the archive.

cheers,

Stan

martindemel · Answer

Hi Hadley and Pete, I like this example, great work!

Some comments to the project and data tables:

* In the final After-Summaries table there are still the Graph Builder scripts from the original After table included - those should be deleted from the projects data table as they will not work and just confuse.
* Be careful wit your profiler. You can see how the FPCs affect the after Glucose shape. However, you miss the information how this change in FPCs will also change the profile of the before glucose profile. So you might want to consider a combined profiler where you can see the before and after at the same time! (see attached file where I combined the summaries information including the formulas)
* In the combined final profiler script in the data table you will see how the FPC1 and FPC3 still affect the After curve, FPC2 does not as the model showed. But the FPC2 will show that there is a big variation in the almost first half of the time recording. But this does not have any affect on the After-curve. This is a great finding,a s we do not need to go back that far in time to predict into the future (if the model holds to be true).
* To get to that profiler is kind of fuzzy and might add questions (or confusion) you might not want to create. But I think we should at least be aware of it. The combined profiler can be also done without FPC2 shown (Reset Factor Grid). This might be better aligned to your story.

My recommendation is to use the combined profiler without FPC2 as final statement, as you see both the before and after and that's what you want to have, if I understood your story correctly.

After - All Summaries.jmp