See also: IRC log
<jluciano> what's eric talking about?
<jluciano> the children's hosp project?
<michel> i2b2
<michel> ericP: simple data structure, no complex relations
<michel> ... they're excited about adding semantics to describe concepts, ericP excited about adding semantics for the relations
<michel> ... patients have a portal into the clinical database
Michel: Open MRS had a demo,
can't really do all things, like have to know patient's
name
... but it is open w meet-ups etc, concept library like
i2b2
<ericP> ericP: there's also MLHIM (Multi-level Health Information Modeling)
Michel: Are there other platforms where we can work on plumbing and concepts?
Eric: All these projects have a
few coders on them
... don't now that i2b2 or Indivio offers serious advantage
over others
<michel> requirements: interface, basic plumbing that can be customized, provide new content
<michel> attending: ericP, michel, scott, bob, joanne, elgar
<jluciano> TIm Lebo
Eric: Stuck a couple of claims in
the paper, when forced to come up w model
... some bag of assertions that need not be elevated into
ontology
... huge amount of info in SNPs, can be condensed into SNP
variation for a person
<mscottm> Another system that I and Matthias looked at back in June: http://www.tolvenhealth.com/ It was being used by Collaborx
Eric: need a SNP expert, maybe
join that with other things
... SNPs appear as XML, we need RDF. What would that look like
in Indivo? Don't quite know
... what is true of all SNPs, don't need to code for individual
patient.
Joanne: Are SNPs relevant for Alzheimer's?
Michel: Value of SNP is hiding in
text: major allele, position, current allele
... info not quite structured in the way we need, but it is in
there.
Eric: Micro-parsing is ahead of
us.
... need to generically represent tests
... there is a whole lot of structure in the patient data about
SNPs that are incomprehensible
Michel: RDF-ize dbSNP, would specify the structure
Eric: We can push this back to Indivo, make an informed plea
Joanne: Paper, still need to use TMO in queries
Eric: trans:, TMO, also strong
desire to force units
... TMO is principled, wrote some OWL to map to units etc
Michel: trans: gets the RDF done,
need a formal mapping trans: to schema relations
... more painful way is to formally analyze every
transformation generates TMO
Eric: There is practical value to have another namespace, like trans:
Scott: Is trans: from bottom-up?
Eric: Yes, nothing special about it
<jluciano> that was Tim Lebo
Eric: trans: is a "to-do"
namespace artifact
... Look at data, look at TMO, ask "How is this supposed to be
expressed?"
... eveantually should not see any trans: in the XSLT
... XML data from EHR --> RDF version of Indivo
Tim: Generally good idea to push the ontology into the converter?
(good question!)
Eric: Logical structure for data,
then that should be how the data are represented
... get rid of the trans as long as the queries do not become
hopelessly erudite
... trans: represents questions people do ask plus what
questions they might want to ask
... as long as questions are intuitive in TMO, then tweak the
XSL
... there becomes less and less trans: in this process
... OWL file is a separate issue, to show fixed units
... tension between that, and TMO
Scott: We have discussed this a
great deal.
... because we are using a language that is amenable to
reasoning then we should use it
(that was Michel!)
Eric: At what level of tooling do
you reduce choices?
... want data normalized *at input*, makes it easier
downstream
Michel: Trivial to normalize units
Eric: Every standard reduces
choices, the cost becomes exponentially harder
... generic federation problems appear after download
Michel: Once it's in RDF and
sparql, well structured, then it't trivial to normalize
... that's what we're looking to do
Eric: Still have to examine each one and deal with it. Q is not whether but when.
Scott: "Standardizing" we require
that people use certain units w. TMO. So how to specify that as
a constraint?
... can't say from TMO that you need certain units?
Eric: Can do some enforcement in
OWL
... Q is What is the space between the patient data and the
TMO?
<jluciano> https://dvcs.w3.org/hg/TMO-Indivo/file/3334734509e9/syntheticPatients/AD_PCHR_1.xml
<mscottm> I lost all audio
<mscottm> got it back again (?)
<mscottm> .rdf files include some CPR - they were produced before using indivo
Eric has missed his lunch.
This is scribe.perl Revision: 1.135 of Date: 2009/03/02 03:52:20 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) No ScribeNick specified. Guessing ScribeNick: Bob Inferring Scribes: Bob WARNING: No "Present: ... " found! Possibly Present: Eric Joanne Joanne_Luciano Michel Scott Tim aaaa aabb attending epichler ericP https jluciano mscottm requirements You can indicate people for the Present list like this: <dbooth> Present: dbooth jonathan mary <dbooth> Present+ amy Regrets: matthias WARNING: No meeting title found! You should specify the meeting title like this: <dbooth> Meeting: Weekly Baking Club Meeting WARNING: No meeting chair found! You should specify the meeting chair like this: <dbooth> Chair: dbooth Got date from IRC log name: 02 Dec 2010 Guessing minutes URL: http://www.w3.org/2010/12/02-hcls2-minutes.html People with action items:[End of scribe.perl diagnostic output]