[R] R and clinical studies

[AJ Rossini]

On Friday 16 March 2007 09:36, Delphine Fontaine wrote:
> Thanks for your answer which was very helpfull. I have another question:
>
> I have read in this document
> (http://cran.r-project.org/doc/manuals/R-intro.pdf) that most of the
> programs written in R are ephemeral and that new releases are not
> always compatible with previous releases. What I would like to know is
> if R functions are already validated and if not, what should we do to
> validate a R function ?

Validation is in the eye of the beholder. 

In particular, for clinical studies, from the corporate or institutional point 
of view, "what we should do to validate an R function" should be answered by 
the local Standard Operating Procedures (SOPs) for "what should we do to 
validate a computer programming language function".   

If you are working with clinical trials as part of a health authority 
submission process, you should have those in place.  

Of course, what you probably are interested in is an approach where you 
qualify R, and validate programs and packages written for R, which might be 
another better approach, in which case the same applies.  Your SOPs should 
apply to both. 

(Now, assuming that you've done a reasonable job on the processes, as per 
Mats' answer, the point is that "R" vs. anything else is a simple red 
herring, as there is nothing in the spirit of the regulations which 
differentiates any of the characteristics of R with any other reasonable 
piece of software, for appropriate definitions of reasonableness).

<digression title="semi-relevant, on SOPs and commercial software">
I should point out that a certain large company I'm familiar with, who uses a 
certain "famous" piece of statistical software for activities perhaps 
described above, can't use the most recent version because of interesting 
issues with its "self qualification" tool, which prevents it from 
self-qualifying the new version on any installation on a certain operating 
system originating near where I used to live, when the previous version of 
the famous software had been installed.  This feature, if not reverted, would 
necessitate total disk wipe of ALL computers requiring qualification running 
this operating system, where the new version of this famous piece of software 
would be installed, if this certain large company wants to follow it's SOPs.   
This is apparently a feature, not a bug, and demonstrates clearly the 
benefits and joys of commercial support when millions of swiss francs of 
licensing fees are involved.
</digression>

I'm not a lawyer, nor am I speaking for any corporation indirectly referenced 
above, nor will I provide sufficient justification to help anyone else take 
any of the statements as a fact.

best,
-tony

blindglobe at gmail.com
Muttenz, Switzerland.
"Commit early,commit often, and commit in a repository from which we can 
easily roll-back your mistakes" (AJR, 4Jan05).
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
Url : https://stat.ethz.ch/pipermail/r-help/attachments/20070317/1c3c0fc4/attachment.bin