15:59:53 <RRSAgent> RRSAgent has joined #hcls
15:59:53 <RRSAgent> logging to https://www.w3.org/2022/02/10-hcls-irc
16:00:01 <dbooth> rrsagent, make logs public
16:00:06 <dbooth> Meeting: FHIR RDF
16:00:28 <dbooth> Chair: David Booth
16:03:15 <dbooth> Topic: Concept URIs
16:04:03 <dbooth> gaurav: Making a list of all codes in FHIR examples for R4 and R5.  Linked from issue 94.
16:04:55 <dbooth> ... Three sets of file: system codes.
16:05:04 <dbooth> ... path_end tells you where it is in the file.
16:06:00 <dbooth> ... "path" is the json path to it.
16:07:31 <dbooth> ... "path_end" is the most specific part of the path (except array index).
16:08:35 <dbooth> ... unique-codes file has only unique codes.
16:10:20 <dbooth> ... Most of these URIs are resolvable.
16:11:24 <dbooth> gaurav: I left out the display value.  I'll add that.
16:12:16 <dbooth> ... Takes about an hour to run.   Works on a dir of JSON files.
16:12:39 <dbooth> eric: Could be a useful tool to the FHIR folks.
16:13:46 <dbooth> eric: The ORIM is the RIM expressed as OWL.  I have something that turns a CDA into Turtle.
16:13:55 <dbooth> ... RIM = Reference Implementation Model
16:14:27 <dbooth> ... RIM ~= v3.   Anything you see that talks about participations, acts, and act relationships comes from the RIM.
16:16:53 <dbooth> ... Next step; write up problem discussion, to make an argument for adding those prefixes.  I didn't see a compositional syntax.
16:18:27 <dbooth> https://www.w3.org/wiki/HCLS/ClinicalObservationsInteroperability/TermInfo
16:20:13 <dbooth> https://www.w3.org/2013/C-CDA/IJ.xml
16:21:07 <dbooth> Topic: FHIR RDF Playground
16:22:07 <dbooth> eric: I make a version to load different profiles.  Can load from any FHIR and get the shex and transformation stuff for that release.
16:23:00 <dbooth> ... Instead of using distant contexts it generates them on the fly from a particular profile.
16:24:44 <dbooth> ... It allows the user to specify which FHIR version to use, but reading the model defs dynamically.
16:25:07 <dbooth> Topic: RDF Lists
16:25:58 <dbooth> jim: Looking at the behavior of specific OWL tools, if rdf:first and rdf:rest are defined as object properties.    In Java OWL API, issues are mainly in the RDF to OWL parser.
16:27:11 <dbooth> ... In example turtle file (books in a library), if you read it into OWL API and write it back out, the properties are maintained, but the instance data is dropped somehow.  Only leaves the initial list node as a bnode, with no other triples about it.
16:27:38 <dbooth> ... If you write the same thing in the OWL functional syntax, OWL API parses it fine, and you can even write out turtle.
16:27:48 <dbooth> ... Neither hermit nor elk complain.
16:28:45 <dbooth> ... Elk doesn't complain, but OWL EL profile does not support anonymous indiviuals at all, making it impossible to use Elk on this kind of data.
16:29:38 <dbooth> ... Will ask the Elk maintainer why.
16:30:03 <dbooth> ... Plan to try other reasoners next, like Pellet.  Will also try some python tools.
16:30:32 <dbooth> ... It's primarily a parser issue that constructs OWL from RDF.
16:33:12 <dbooth> Topic: FHIR RDF examples knowledge graph
16:35:01 <dbooth> dagmar: Found a nice graph visualizatoin tool.  It shows nodes in different colors for differnet types.
16:36:05 <dbooth> ... LOINC code is far away (in the graph) from the pt node.
16:36:26 <dbooth> ... Maybe eliminate some of the blank nodes to make the knowledge more condensed.
16:36:50 <dbooth> ... Gartner predicts 85% of data in graphs by 2025.
16:37:42 <dbooth> ... Thanks for OpenEHR and CIMI links.
16:38:16 <dbooth> ... What kind of KG should be stored about the patient?
16:39:20 <dbooth> eric: One value is if you filter out things like fhir:value, you can look at constellation of continuty of principles in this dataset?
16:39:31 <dbooth> ... Lots of practitioner 005.
16:43:42 <dbooth> david: suggest think about what you want to do with the data, rather than trying to make an "ideal" representation.
16:44:34 <dbooth> dagmar: Want to make it easier for people looking at it to understand.
16:47:14 <dbooth> david: Suggest looking at FHIR Shorthand (FSH) https://github.com/HL7/fhir-shorthand
16:48:19 <dbooth> Topic: FHIR RDF playground
16:48:59 <dbooth> (Eric does demo of loading different FHIR version into the playground)
16:49:47 <dbooth> https://ericprud.github.io/fhircat-playground/playground/
16:49:55 <dbooth> click the R4 button (which it loads by default)
16:50:05 <dbooth> you can also drag and drop a zip or the URL of a zip (typically a link from the downloads page)
16:53:29 <dbooth> Present: EricP, David Booth, Dagmar, Jim Balhoff, Gaurav Vaidya
16:54:11 <dbooth> rrsagent, draft minutes
16:54:11 <RRSAgent> I have made the request to generate https://www.w3.org/2022/02/10-hcls-minutes.html dbooth
17:08:38 <dbooth> rrsagent, make logs public
17:08:42 <dbooth> rrsagent, draft minutes
17:08:42 <RRSAgent> I have made the request to generate https://www.w3.org/2022/02/10-hcls-minutes.html dbooth