W3C home > Mailing lists > Public > public-hcls-dse@w3.org > January to February 2007

RE: draft template for Stephen's clinical data

From: Eric Neumann <eneumann@teranode.com>
Date: Mon, 26 Feb 2007 10:38:18 -0500
Message-ID: <A3970D83EC72E84B8D2C2400CD6F0B9FE9B2D4@MI8NYCMAIL16.Mi8.com>
To: "Forsberg, Kerstin L" <Kerstin.L.Forsberg@astrazeneca.com>, public-hcls-dse@w3.org
Kerstin,

Here is the Time Ontology page for everyone's reference:  http://www.w3.org/TR/owl-time/
basically it is composed of time instances and durations (intervals with a beginning instant and an end instant)

I think we can apply it to the following aspects:

- Period (Interval) of study
- Birthdate (Instant) of patient
- Age (Interval) of patient (determined from from birthdate and study beginning)
- Interval of treatment (or each arm for dose changes)
- Date (Instant) of visit
- Interval of AE
- Duration (Interval) of body position or activity (context) 
- any others?

As for CDISC vocab, I see mainly predicates (e.g., SDTM columns) and (e.g., MedDRA codes); where we will differ from the standard SDTM is in the possible use of blank nodes (e.g., context) and URIs. Is this a fair assessment? 

Lets plan to discuss this on friday...

Eric

-----Original Message-----
From: public-hcls-dse-request@w3.org on behalf of Forsberg, Kerstin L
Sent: Mon 2/19/2007 3:39 AM
To: public-hcls-dse@w3.org
Subject: Re: draft template for Stephen's clinical data
 

Hi Eric,
the first draft for the three kinds of observation data do raise a number of question in my head.

We assume a quite large CDISC vocabulary (http://www.cdisc.org/sdtm/vocab) that does not exist today. I think it is important that we think through how such a vocabulary should be constructed. Should we keep it close to the way CDISC today define their tabulation model, general observation classes and domain specific variable names that implies a flat file structure approach? Could be a good alternative if we would like to make our case stronger in our discussions with CDISC. Or, should we construct one or more ontologies that applies a more "semantic web approach" of observations as evolvable information resources instead? 

The integration of the time owl ontology makes good sense but I do get a bit confused on how it should be applied.

As discussed in the last TC, I think it is important that we include examples showing the value of having assigned URI:s for annotating clinical data, such as Adverse Event encoding using MedDRA codes. Instead of only using strings of texts.

With regards to the first question about the cdisc vocabulary / ontologies I will try to exemplify the two different options for our next TC. However, I may have to decline that TC on Friday due to a conflicting meeting.

Kind regards
Kersin
Received on Monday, 26 February 2007 15:41:52 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 14:11:53 GMT