Re: DSE SDTM Note

Kerstin,

Thanks for taking the time to review and give suggestions... I will
add them into the draft and have it on the web for tomorrow's
sessions.

cheers,
Eric


On 11/7/07, Forsberg, Kerstin L <Kerstin.L.Forsberg@astrazeneca.com> wrote:
> Hi Eric,
> good to see that you have put together a W3C Note on SDTM from our collaborative work on the DSE wiki. I have some comments on the draft version of it:
>
> *       A graph extensible form (SDGM?) in RDF that you propose would be very useful for clinical data with many different types of associations such as biomarker data and genotyping from subjects. Important to notice is however that the current tabulated form do have an extension mechanism called Supplemented Qualifiers. A SUPPQUAL dataset consist of supplementary qualifiers extending the predefined and permissible variables for record qualifiers in the different SDTM domains. SDTM also has also a way to relate records from different domains. For example relating the Pharmacokinetic Parameters to their Concentrations. However, CDISC do not provide a mechanism to define what qualifiers are required for each type of observation.
>
> *       The scenario talks about a specific set of genetic diagnostic data while the examples is about more general type of clinical data.
>
> *       In the examples you assume a namespaces called http://www.cdisc.org/sdtm/vocab that would have to include classes and properties such as Study and Subject (i.e. either domain entities or records describing these entities?), VSTest (i.e. record structures),  StandingDiastolic_BP (i.e type of records), obs_units and obs_values (i.e. fields in records). Based on our recent email-list discussions I think this is an example of the kind of mixture of domain and record level Alan, Vipul and I are concerned about.
>
> *       I propose that you take away the section for "Bundling Diastolic_BP, Systolic_BP, and Pulse together under one observation" as this blurs the concept of an unique identified instance of an observations.
>
> *       In the section "Discussion regarding CDISC SDTM Code lists" I think you should take away the first part with the attached XML files from NCI caDSR as CDISC now instead publish their controlled terminologies as a Excel matrix incorporating the terms and definitions from NCI Thesaurus. I propose that you reference the slide that I have put together that describes these relationships see slide 4 in http://www.w3.org/2007/11/NCIt_and_CDISC_CT.ppt
>
> *       Glad to see that you picked up my proposal regarding "observation types ontologies" as part of the section "Discussion regarding CDISC SDTM Code lists". However, I think the proposal should be better aligned with the outcome of our discussions on domain vs. record level.
>
> *       In the examples you have Sex as a attribute of Subject. That is in my mind a second step and depending on the types of studies and projects. Some of the observations on a subject are considered to be fixed characteristics of the subjects. Sex and Race are the obvious examples, while in others are much more project and study specific such as smoking habits.
>
> Kind Regards
> Kerstin
> *
>

Received on Wednesday, 7 November 2007 15:26:06 UTC