search

pharmaR / riskassessment

Collaborative Deployment: https://app.pharmar.org/riskassessment/ Risk Assessment Demo App: https://rinpharma.shinyapps.io/riskassessment
https://pharmar.github.io/riskassessment/
Other
101 stars 29 forks source link

Consider decision tree flow chart #601

Open AARON-CLARK opened 1 year ago

AARON-CLARK commented 1 year ago

Per feedback from useR@ 2023 conf in Basel, @Xyarz reported the desire for this issue. Did I describe it correctly Lars?

I can confirm, there are all sorts of pharma orgs using these tree diagrams to make their GxP inclusion decisions using automated quality checks, especially in the 2022 case studies. Nodes would have to absolutely be set up using the config file. I've pasted some examples below!

image

image

image

Xyarz commented 1 year ago

Yes! Thanks @AARON-CLARK. The ask was regarding a tree based algorithmic decision.

AARON-CLARK commented 1 year ago

User feedback

So at the end of the day, how do we help users understand how the decision automation rule's code was evaluated? Right now, it's very "black boxie".

  1. If a decision isn't made, that means all of the rules ran and nothing returned TRUE. I'm not sure what we should display to reflect this. Doing nothing is okay for now, but what if I expected something to happen (like in the example above) but don't have a way to verify the results? Do you have any thoughts? In an ideal world, It'd be nice to show both the .x value and the resulting TRUE/FALSE for each condition (rule), but that's difficult to display in the Upload Summary Table. Perhaps we could tackle it in a future PR.

Originally posted by @AARON-CLARK in https://github.com/pharmaR/riskassessment/pull/632#pullrequestreview-1646627748

User feedback

This is a big question that I am still struggling with. I suppose we should at least store dummy assessment data and verify that the rules will actually run before allowing a user to add them to the database. I will think more on how to display what occurred when a package is uploaded.

Originally posted by @Jeff-Thompson12 in https://github.com/pharmaR/riskassessment/issues/632#issuecomment-1737621898