Re: [TMO] patient record normalization

* Kerstin Forsberg <kerstin.l.forsberg@gmail.com> [2010-09-12 18:00+0200]
> Eric, Chime, Michel, Alan et al.'
> 
> This is a truly important discussion for all of us that work on
> patient records / clinical data.
> 
> Conclusions and recommendations on the issue of units for numerical
> data values given semantic web standards and linked data principles
> from the combined intelligencia of HCLS, TMO and IAO would be very
> much appreciated. A parallel issue is domains of enumerated values for
> coded data.

Here are the rules that I'd apply:

For scenarios with lots of manual user intput, allow unit flexibility
and express the values and units in a popular vocabulary like muo.
This puts the burden on the computer instead of the person (consider
CSS, which allows input in px, em, in, cm, mm, anstroms, light-years,
etc.)

For the bits of information where you don't intend to minimize
heterogeneity, either because it's socially impractical or because
it leads to combinatorial explosions of term names, use the popular
value/units vocabulary

For standard representation of information where you wish to eliminate
heterogeneity (as I'm trying to do with blood pressure and some other
vitals), express the data with a simple data structure (no units
variant) and add OWL to the ontology to map that data structure to the
value/units vocabulary.


> Regards
> Kerstin

-- 
-ericP

Received on Sunday, 12 September 2010 16:32:09 UTC