See also: IRC log
Ken: Want to find ways to combine MDMI with RDF work
<scribe> ACTION: David to schedule discussion on comining MDMI and RDF efforts [recorded in http://www.w3.org/2017/04/04-hcls-minutes.html#action01]
<trackbot> 'David' is an ambiguous username. Please try a different identifier, such as family name or username (e.g., dderour, dhansen2, dnewman, dshotton).
eric: Status codes have semantic
relationships that are not explicit. Would be helpful to make
them explicit in OWL.
... Then inference engines would understand them.
... We want that hierarchy written not only in RDF, but in the
FHIR spec, so that the RDF will stay up to date.
ken: We noticed instances where
the state machine model in the VA was wrong, relative to codes.
So this would help.
... Is there one ont or status code set across different FHIR
resources?
eric: To the extent that people did not want to reinvent the wheel, they are harmonized across resources.
rob: Not required to be harmonized, but there is an effort to do so.
dbooth: If the same code is used in different resources, are we guaranteed that it has the same meaning?
rob: Yes if they have the same
coding system.
... The value set lists where they are used: http://build.fhir.org/valueset-observation-status.html
matt: There are some lexical codes that are the same, but with different meaning, such as in http://build.fhir.org/valueset-diagnostic-report-status.html But they have differrent namespaces, so they are distinguished that way.
dbooth: What if we have codes in different systems that are logically equivalent?
matt: Definitions are 95% identical, but not 100%.
dbooth: How much added benefit would there be for expressing that equivalence?
eric: Order and fulfillment pipelines would have use cases to benefit from equivalence. But clinical questions would probably need explicit status anyway.
dbooth: I could imagine wanting to do a query using only data with status=final.
eric: Harmonizing across resources would be much harder than getting the individual hierarchies expressed in RDF.
dbooth: How should these mini-ontologies be organized? One file per mini-ont? One file combining the min-onts? Part of fhir.rdf?
eric: one advantage to one file
per mini-ont is that the ones that have been harmonized can
then be moved into a different place, such as fhir.rdf
... If we want to do inference we probably want them to be OWL
individuals.
... One file per mini-ont allows us to mimic the existing FHIR
structuring.
AGREED: Generate one mini-ont file per valueset
<scribe> ACTION: eric to request indenting to indicate hierarchy use explicit symbol, such as arrow in http://build.fhir.org/valueset-observation-status.html [recorded in http://www.w3.org/2017/04/04-hcls-minutes.html#action02]
<trackbot> Created ACTION-80 - Request indenting to indicate hierarchy use explicit symbol, such as arrow in http://build.fhir.org/valueset-observation-status.html [on Eric Prud'hommeaux - due 2017-04-11].
<scribe> ACTION: Rob to check on the resolution of final/amended codes (whether amended is an end state) http://build.fhir.org/valueset-observation-status.html [recorded in http://www.w3.org/2017/04/04-hcls-minutes.html#action03]
<trackbot> Created ACTION-81 - Check on the resolution of final/amended codes (whether amended is an end state) http://build.fhir.org/valueset-observation-status.html [on Rob Frost - due 2017-04-11].
<scribe> ACTION: Rob to work with Harold on implementing mini-ont generation [recorded in http://www.w3.org/2017/04/04-hcls-minutes.html#action04]
<trackbot> Created ACTION-82 - Work with harold on implementing mini-ont generation [on Rob Frost - due 2017-04-11].
dbooth: Are there other open source projects that we should invite to our upcoming discussion, about coming/coordinating efforts?
ken: You deal with RDF space, we deal with UML space. There are a lot of people using open source tooling for modeling information models.
dbooth: Shared goal is gaining interop using open source.
ken: Like to see what we're doing with Linked Data.
dbooth: Linked Data is RDF. Another name for it.
ken: Maybe OpenEHR?
dbooth: Who should we ask?
rob: Tom Beale (best to ask), Heather Leslie, Ian McNichol
ken: In OMG we're about to start working on v2 of the MDMI standard, and Elisa Kendall will help. We would welcome participation.
eric: Can we make use of the
models, by virtue of harmonizing with RDF? The BRIDGE was
written in UML, then put in RDF and I was able to use it.
... Cecil Lynch probably wasn't thinking about what would be
needed, but I was able to directly use his work.
... A success would be if all the CIMI terms could be
harmonized, so two people write bodyMass, we can know that
they;re the same. OpenEHR could do then instead of using
numeric IDs.
... Their ID system looks like JSON-Path at the moment.
... That could be an objective.
ADJOURNED
This is scribe.perl Revision: 1.152 of Date: 2017/02/06 11:04:15 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: Irssi_ISO8601_Log_Text_Format (score 1.00) WARNING: No "Present: ... " found! You can indicate people for the Present list like this: <dbooth> Present: dbooth jonathan mary <dbooth> Present+ amy <amy> Present+ No ScribeNick specified. Guessing ScribeNick: dbooth Inferring Scribes: dbooth Found Date: 04 Apr 2017 Guessing minutes URL: http://www.w3.org/2017/04/04-hcls-minutes.html People with action items: david eric rob[End of scribe.perl diagnostic output]