Update analyse() to support data.frame input

[gowerc]

Apologies I appreciate this request has previously been discussed the last time we met during which we decided this is something we wouldn't include into the package.

That being said I was talking to a former colleague who mentioned that at least 2 other companies he's worked at also use an equivalent to an ADMI dataset when handling multiply-imputed values which is then submitted to the HA's as part of their submissions.

Based on this I am wondering if this is worth re-considering as the impression I am now getting is that this workflow is more common than not across the broader industry.

[danielinteractive]

Hi @gowerc and @luwidmer ,

I just come across this topic from another company's project side, where they follow exactly this workflow. Now we wonder how to make that work with rbmi.

1. Do you or others have any prototype code to make this work with rbmi? (I appreciate we could kind of hack a solution relatively easily together, just wondering if there is something that can still then use the analyse() call)
2. Would you be open to an extension of rbmi to support this? I could potentially help with this.

[tobiasmuetze]

The {rbmiUtils} package from @bailliem adresses this.

[danielinteractive]

Ah that is great! I just checked the GitHub pages site etc. but was not 100% sure what is the best entry point for this use case, is there a dedicated function or vignette for it maybe? Is it https://openpharma.github.io/rbmiUtils/articles/pipeline.html#analyse-each-imputed-dataset ?

[luwidmer]

I think so, but @bailliem would know better given it's his package 😊

[bailliem]

Ah that is great! I just checked the GitHub pages site etc. but was not 100% sure what is the best entry point for this use case, is there a dedicated function or vignette for it maybe? Is it https://openpharma.github.io/rbmiUtils/articles/pipeline.html#analyse-each-imputed-dataset ?

Hi @danielinteractive , the use case (beyond company norms) that might be helpful is transforming on the imputed data sets. This is an example of it here https://openpharma.github.io/rbmiUtils/articles/deriving-endpoints.html

Hope that helps make it clearer.

[bailliem]

@danielinteractive , and just to add. The rbmiUtils package is mainly motivated to abstract away some of the awkard things you would have to do to make rbmi work to fit into this use case. Saving imputed data sets as ADMI is not something I think we have fully aligned on yet.

[danielinteractive]

Thanks a lot @tobiasmuetze , @bailliem , @luwidmer 🫶 super helpful!