I'm unsure if you can create these on your own, the ones I'm aware of are built in to NNMi, but think this is the concept you're looking for....

In NNMi, there is functionality where (just an example) if an interface goes down, it will trigger an NA Interface diagnostic.  

Think they are called event action rules or something similar.  

If you have the NNMi / NA integration enabled and no one has turned these off, then you should be able to log into NA, run a task report and look for tasks run by the ID that is used for the integration.  They'll be diagnostic tasks.  

Note: don't be surprised if they are skipped (bug, there is a fix - not related to this topic)  

But, back to your question, that would be the way I could see this working through NNMi.  However, I am unsure if this is doable today.  You may want to submit an Idea Exchange.  

Other option might be to use Operations Orchestration (if you have this within your company) - I could see this being a possibility.  Or perhaps through your ticketing system.  Example - NNMi sees high utilization and it triggers a ticket within whatever system you are using.  When your ticketing system gets the request, there is logic that would then use the NA API, connect to NA, run the diagnostic you want and capture the results and place those details within the ticket (so the assigned support person doesn't have to do that).  

Hope this helps.

-Chris