All Minutes

Meeting minutes

ElementDefinition summary flags #133

tim: Summary info is for searching for data and not wanting to provide everything.
… Not in the RDF ont right now, but it's in the element definition. Should it be in the ont?

eric: ShEx schema has a def for StructureDefinition, which means that any StructureDef (in principle) has an RDF representation. We exlude all the metadata when we're building ont.
… I.e., we din't put much in there -- not the metadata.
… We could, or we could write a script that emits the StructureDefs, to have access to everything in them.

tim: It's actually ElementDefinition

eric: Could emit it as an RDF representation of the ElementDef, as an instance of it, e.g., for Patient

structure def for all resources: https://www.hl7.org/fhir/structuredefinition-examples.html

https://www.hl7.org/fhir/patient.profile.json.html

eric: If this were dumped out as RDF, you'd have the info you need.
… But you'd need to do som estring manipulation.
… Adding a fhir:link would help, but it would have to point back to the shex, because in some resources it will have isSummary true and in some it will be false.

eric: COuld add it to the fhir ont.

https://www.hl7.org/fhir/elementdefinition.html

eric: Need to talk about the use of "identifier" in the context of the patient resource.

dbooth: Add this meta data to a separate ont file?

eric: It could require the regular fhir ont.

eric: We already have shapes for structure defs. Could have patient.resource.ttl, having some element defs w a string w patient.identifier, and you'd have to know that that is the same as patient.identifier in your resource.
… Will you perform those string operations anyway? if not, and we put in a fhir:link pointing back to the ont, and instead of using bnodes for the property constraints, we could have named nodes.
… Could also point at the shex and use shapepath to identify the spot in shex.
… Other possibility would be do string manip.

tim: Fine w the strcutre def approach, as long as there is something tying it to class patient, that should be enough, if we have the RDF structure def of patient.

ACTION: Tim to come up w concrete proposal

HL7 working group mtgs

https://tinyurl.com/fhirrdf-2023-sep

dbooth: Some jira tickets marked as reverted, "reconsider for R6".

rob: Need to redisposition it, as completed, currently in R5, mark as preapplied, and ask ITS to approve it.

ACTION: DBooth to ask Paul Knapp about zoom

Playground

houcemeddine: Need the variable for turtle.
… Also, getting 404 for claim URL.
… Wrong URL: http://hl7.org/fhir/shape/Claim
… Right URL: https://build.fhir.org/claim.shex

rob: Also those URLs are not going to the same release. One is the CI build; the other is the official release.

eric: FHIR use case conventions, but the resource system doeesn't.
… Capital letter things like Observation ...

houcemeddine: If you lowercase the name, then the file is there.

eric: Maybe it's just the way deepak is generating them.

eric: It's a very flat hierarchy. Might be easier to get case sensitive URLs in a subdirectory.

houcemeddine: Also another problem in converting JSON to Turtle. Fails for medication in R5.

eric: There was a skew between the schema and the data.
… Probably need to get the newest examples.

ADJOURNED

Meeting minutes

Update on FHIR RDF

DBooth: (Reviewed slides at https://tinyurl.com/fhirrdf-2023-sep )

Corrected reverted R5 decisions

dbooth: Several FHIR RDF changes were completed, voted, passed and PRs merged, but we apparently neglected to actually close the jira tickets.
… So when Grahame and others were cleaning up unfinished tickets, they thought that these were unfinished, and they reverted the decisions.
… But they were in fact already fully done and merged and are in the official R5 spec.

Paul: We can fix that
… We are now putting back the disposition and the vote that was there before, and closing the items as they needed to be closed.

dbooth: here are the issues:

https://jira.hl7.org/browse/FHIR-39074

https://jira.hl7.org/browse/FHIR-39075

https://jira.hl7.org/browse/FHIR-39076

https://jira.hl7.org/browse/FHIR-39073

dbooth: This issue is different, because it was moved to THO: https://jira.hl7.org/browse/FHIR-37960

paul: This issue should stay open until THO has resolved it.

Approval of minutes

dbooth: Motion to approve.

eric: Second.

APPROVED: 5-0-0

https://www.w3.org/2023/02/20-hcls-minutes.html

https://www.w3.org/2023/02/22-hcls-minutes.html

https://www.w3.org/2023/03/02-hcls-minutes.html

https://www.w3.org/2023/03/09-hcls-minutes.html

https://www.w3.org/2023/03/16-hcls-minutes.html

https://www.w3.org/2023/03/23-hcls-minutes.html

https://www.w3.org/2023/03/30-hcls-minutes.html

https://www.w3.org/2023/04/06-hcls-minutes.html

https://www.w3.org/2023/04/13-hcls-minutes.html

https://www.w3.org/2023/04/20-hcls-minutes.html

https://www.w3.org/2023/04/27-hcls-minutes.html

https://www.w3.org/2023/05/04-hcls-minutes.html

https://www.w3.org/2023/05/11-hcls-minutes.html

https://www.w3.org/2023/05/18-hcls-minutes.html

https://www.w3.org/2023/05/25-hcls-minutes.html

https://www.w3.org/2023/06/01-hcls-minutes.html

https://www.w3.org/2023/06/08-hcls-minutes.html

https://www.w3.org/2023/06/22-hcls-minutes.html

https://www.w3.org/2023/06/29-hcls-minutes.html

https://www.w3.org/2023/07/06-hcls-minutes.html

https://www.w3.org/2023/07/13-hcls-minutes.html

https://www.w3.org/2023/07/20-hcls-minutes.html

https://www.w3.org/2023/07/27-hcls-minutes.html

https://www.w3.org/2023/08/03-hcls-minutes.html

https://www.w3.org/2023/08/10-hcls-minutes.html

https://www.w3.org/2023/08/17-hcls-minutes.html

https://www.w3.org/2023/08/24-hcls-minutes.html

https://www.w3.org/2023/08/31-hcls-minutes.html

ADJOURNED

- DRAFT -

FHIR RDF

05 Oct 2023

Attendees

Summary of Action Items

Meeting minutes

Turtle parser/serializer

tim: Next I want to work on converting strings to booleans, but need a quicker way to test.

dbooth: Can meet right after this call, with Deepak, to discuss.
… Any other questions about what should be done?

eric: Root nodes in R5 are now bnodes instead of IRIs.

jim: I think it was because in R4 it had BASE setting.

dbooth: Wny would lack of explicit BASE prevent it from getting an IRI? Wouldn't a default BASE be used?

eric: In theory yes, but if nothing sets the default base then the code does a bnode.

https://hl7.org/fhir/R4/observation-example.ttl.html

versus this: http://build.fhir.org/observation-example.ttl.html

jim: https://github.com/hapifhir/org.hl7.fhir.core/blob/master/org.hl7.fhir.r5/src/main/java/org/hl7/fhir/r5/elementmodel/TurtleParser.java#L405

https://github.com/hapifhir/org.hl7.fhir.core/blob/master/org.hl7.fhir.r5/src/main/java/org/hl7/fhir/r5/elementmodel/TurtleParser.java#L320

dbooth: Any questions about how this should be handled?

jim: I have a SPARQL tool to skolemize all blank nodes into IRIs using UUIDs.
… I'd like to incorporate that if we continue in this direction?

eric: If you can emulate Harold's SNOMED example, that would be good. I was planning on showing something like that at AMIA.
… Need to have 3.5 hour video recorded by the 27th.

dbooth: Is it clear that we'd like to gen IRIs if we can?

jim: Yes, though if we skolemize it doesn't matter so much.

eric: These should be named nodes, because the REST API allows you to address them.
… Want to encourage people to do that.

dbooth: Will there be an issue of where the examples are hosted vs where the base needs to be?

eric: No, because they're all broken already.
… Would be great to fix this in the publication spec anyway, but reluctant to attack it.

dbooth: Should we show the BASE in the example?

eric: Not helpful. Base statement would have to be changed when you move the data, and a base comment would be more likely to cause confustion than help.

Concept IRIs

gaurav: CURIES have a prefix in the code. Wikidata has this same problem.
… Joe Flack from TIMS contacted me. They're trying to figure out what to do.
… He added a couple of proposals.

w3c/hcls-fhir-rdf#127

eric: Nothing says the code will be unique across all code systems.

gaurav: I recommend only having the code - -no prefix.

eric: I think stripping out the prefix effectively creates a comment field.

jim: If we allow CURIEs then we don't need system at all.

jim: Suggest we say that you can't put CURIE in a code field. You need to use only the code.

AGREED: you can't put CURIE in a code field. You need to use only the code.

ACTION: DBooth to create jira ticket to add this clarification to rdf.html page

ACTION: Eric to reach out to CSIRO group about CURIES and our FHIR RDF decision

gaurav: Will this affect their triples?

jim: Those are hard coded to use the right RDF namespace.

tim: Is that because someone is putting it into the system and code, but need a process to turn it into RDF links.

jim: Ideally there should be code that pulls the IRI stem and converts the JSON to FHIR RDF using them.

Jira issues

gaurav: Currently need to login to gitlab. But cannot edit a file.
… Cannot open in web UID.

rob: I always do it in a local copy.

gaurav: I tried pushing changes, but it says I don't have permission.

rob: They make a new supposedly simpler procedure. Ask Jessica Bota how to do it.

CURIEs

Eric: ADHRC issue text:

https://hackmd.io/xLPfK2hxS3mhdIIfKQu1xA

gaurav: On vac end of next week for two weeks.

ADJOURNED

Meeting minutes

FHIR Turtle parser

tim: Coming along

Element definition summary flags

w3c/hcls-fhir-rdf#133

tim: Resource properties can be designated for inclusion in the summary of the resource, e.g., pt name.
… Should it go into the ont?
… Could add an annotation to the property restriction. But it makes the turtle more verbose, using OWL.

tim: The first way I proposed doesn't work. But the latest does. A kind of reification.
… If we want to have another turtle file that has this metadata, IDK how to reference the property restrictions in another OWL file.

eric: One thing to check: Whether we have machinery to emit the RDF representaiton of the profiles.
… In principle, the same RDF-ization rules applies to the profiles as the profile instances.
… If so, then we'll already have that info in it. That might be the shortest path to a solution.

tim: This summary flag is actually in the profile, but I don't see the connection between the pt profile and the pt class in the ont.

dbooth: That sounds like an omission.

eric: We added fhir:link other places where we needed to connect things.
… Wonder if the profile name in the RDF-ified version should be the same as the class, or link to the class somehow.
… Probably no glue between the element definitions in the profile and the class constraints in the OWL class.

tim: Original use case is if I have a set of patients in a list, and I only want to show the summary properties.
… But I think that's doable w the structure def as long as it is connected to the class.
… I think a link from the profile to the class would be enough, as an alternative to annotating the properties in the class.

eric: Another alternative: put the annotations onto the shex representation.

dbooth: Feels like it should be part of the profile.

eric: Code in HAPI only works w the resources, pre-baked representations of the resources -- reflection-like clsses. Can't handle things that don't already have a java object.

tim: Makes me think in general: What is the relationship of a profile and the class?
… A profile in the ont is a piece of info, but a pt class represents a pt record.

eric: Playground code turns FHIR JSON objects into RDF. Maybe that would be a place to start?

dbooth: Are we generating profiles in RDF?

tim: I don't see examples of element definitions.
… But I see examples of structure defs that are in Turtle.

eric: That suggests that there's code for parsing structure defs of structure defs.

tim: Yes, but I don't see examples of element defs.
… Actually I see some, but not Turtle versions.
… Is there some part of an element def that could point to a class in the FHIR ont.

rob: Could put an extension in there.

tim: There are XML representations of the pt record and the structure def of pt. Are they different? If so, might want to connect them in XML also.

eric: If I have a profile of a person, structure def and instance of a person (Bob), and resource type Patient. Is that the glue you mean?

tim: Not an instance of a resource. In RDF we have a pt class, and an instance of a structure def that is for a pt.

eric: I don't think there's an analog to the pt class in JSON or XML.

dbooth: Agreed.

eric: What should we call that relationship between the structure def?

dbooth: It's a relationship, we could make up a name for it.

eric: Resource can have 4 defined-bys: structure def, shex, XML schema, JSON schema.
… Then there's some other relationship that points to the OWL class.

eric: profile-validated-by relationship could be on the shex, XML schema or JSON schema.
… PROV prov:derived-from ont can say where this thing came from. Could point from resource node back to the structure def.

tim: I've been working on a linked data browser, and I want to list patients, but only the summary properties.

dbooth: Propose that tim try it out, hypothesize a link.
… and see how it works out. If it works out well, propose it as our solution, then we'll decide where to generate it.

Eric's linked data browser

eric: Right now only works with TriG

dbooth: Running all in the browser? Eric: Yes.

eric: CORS causes lots of errors. Works best on the same host.

eric: example: https://fdpcloud.org/clickable/client/RenderClickableLd?browse=https%3A%2F%2Ffdpcloud.org%2Fsites%2Fgithub%2FStaticFDP%2Fstatic-fdp%2FFairDataPoint.ttl
… It helps find things that are broken.

tim: I've been working on an EHR skeleton, all client-side JS, but pulling in the entire ont and can browse all of the instances of all the example resource instances.

eric: Can you navigate from patient to observations?

tim: Yes.
… THere's a typescript library if you're using the FHIR resource. But if you have a pt type, how can you iterate through them and list the properties of each type without reflection? But I can do that in RDF.

ADJOURNED

ACTION: Tim to demo his browser next week

Meeting minutes

Demo of FHIR RDF app by Tim Prudhomme

tim: All client-side JS. Loading the whole FIHR ont.
… Also loading all the examples, which slows it down. 13" Macbook.

jim: There's a JS lib to load all into an hdt file -- Header Dictionary Triples, compressed.

tim; I've used rdflib JS library, but right now doing this all in clojureScript.
… Using Datalog queries, asami
… Very similar to SPARQL.
… This app has very little code that is specific to FHIR. Almost all info is in the ont.
… Using W5 RDF to organize the data.
… It gives me quite a lot to work with.
… Select a class, that shows instances. Then select a patient instance, it shows info about that instance.
… Can also do that for ingredient.
… Right now the list of pt instances are only ID, name and DOB. But could use summary info if it were available.
… In pt instance it shows cardinality.
… Plan to make it editable.

eric: Editing could issue a PR to github to update the example!

tim: Can use the domain and range to determine the type, when a property is added.
… Not using shex yet.
… Interested in using it for adding new relationshps.
… The RDF and shex gives you everything you need to do most CRUD app work.
… I think only the W5 groupings required FHIR-specific code.

eric: Familiar w FHIR REST API?

tim: yes. THought about manip this data w/o making too many trips to the server.
… Thought about how to make it scale to a real app. Q: How are you persisting the data? Could use triplestore, but don't need to. Could use FHIR API.

dbooth: This could be a useful playground tool for browsing FHIR examples.

jim: Could you select the examples you want to load?

tim: Yes, could.
… This is using clojureScript and datalog.
… In terms of making this open source, JS might be easier for others.
… I could maybe turn this into something that's being used w real healthcare data. Maybe publish a paper about it.
… Main idea for me is to avoid putting business logic into code. Put that into RDF, load it as data.

dbooth: Current plans?

tim: Keep working on the interface. Figure out how to load resources dynamically.

dbooth: Who would use the app and how?

tim: Not sure. What an EHR would provide?

dbooth: Gives you very flexible data exploration. Might be good for research purposes.

tim: Might have a little analytics model.

dbooth: At least show counts.

eric: Want to pick your brain on rendering nested objects

dbooth: Might be a good topic for a Yosemite Project webinar.

dbooth: How to handle non-FHIR data merged in?

eric: Could have SNOMED or other ont.
… If you had env data, and ER data, and you tried to figure out from admission and diagnostics and correlation w pollutants, you might be able to say "When Sulfer dioxide is emitted, these people show up in ER a few days later".
… A lot of it would be temporal and GIS data.

tim: How to display it?

dbooth: Does datacube have diplay info?

eric: It has catalogs. Doesn't do what CSV-on-web of giving semantics. Punted on that.

https://www.w3.org/TR/vocab-data-cube/

eric: DDI gives you more. https://ddialliance.org/Specification/RDF

ShEx for driving UI

(demo by eric)

eric; Shex annotations help in form generation.
… It generates a form using that info.
… .Then when you enter data, it checks it against the shex.

eric: You can also separate out the layout annotations, using shexPath to express them separately.

ADJOURNED

Meeting minutes

ACTION: DBooth to put full zoom info on RDF mtg page, for dial-in

Introductions

Erich: dep of informatics, working on RDF 15 years, working on software for teachers, bridging to clinical . Stonybrook Univ.

Relating imaging data to clinical info

erich: Tagging onto OMOP group and, diacon group. Trying to integrate spacial info, and make heatmaps.
… I'm using RDF, but want to walk it to clinical info also.
… Any work done w diacom or OMOP?

eric: Some of the same people worked on an OMOP mapping a while ago.
… Didn't do anything between images and OMOP. Not aware of images to OMOP. Maybe Scott Marshall?

erich: Imaging and clinical domains seem to be separated. Want to bring them together.
… OMOP is tyring to use Daicom verbiage.
… Have annotations by the files are usable, but cannot bring up 3M image iunless there's spacial info.
… Cannot decode the image and put it on the screen quickly. Need spacial info.
… Same applies to deep learning pipelines.

eric: Write-up of baby use case for clinical setting?

erich: Not aware. Also work w cancer imaging archive. They get a lot of images submitted, and de-ided clincial info. People now want both.
… Diacom is noce, but unless you set standards of what should be included, you get a lot of empty boxes.
… Noe of these groups are bridgeing each other.

dbooth: Need to get folks together to duscuss?

erich: David Kume said thatere;s some diacom, but no official mapping. Series of XML docs, you need to extract info from them.
… FHIR is better, OMOP has CSV which is better. Need to convince people that this is needed.
… Easier to build something to show them the value, to get them hungry to discuss.
… Otherwise OMOP group has 100 people.
… Any image could have 1M segmentations, with lots of info associated.
… Cannot use standard RDF DB -- billions of triples.
… Need to break into separate storage, then need to get clinical date into it.
… In semweb, lots of greag components, but nobody knows what it becomes yet.

tim: little familar w diacom. Thought diacom modalities didn't seem they had depth. Have you looked at relevant FHIR resources? Visionoing more detailed modeling of those modalidties in FHIR?

erich: Looking for low hanging fruite. Collected a lot of COVID imagery 80-100M radiology images.
… Wrote software to extract diacom data and RDF-ized it. Loaded into virtuoso to query.
… Diacom allows linking by accession IDs, dates, etc.
… Brought into relational data laek, but how to connect them back? Some terms connect, but not all.
… Stonybrook purchansing scanners to scan. Say we'll need to decide what to collect on patients, etc. Otherwise you just have a big [pile of images.

dbooth: Maybe show what you have so far?

erich: Could show soon as use case. ONe odball thing: images range from 300M to 3GB, and my computer gets chewed up.
… Having the pipeline data from deep learning, in central RDF DB is not scalabl, virtuoso w TB of RAm.
… With a lot of these files, putting into central store, you're doing spacial indexing, but considering it all part of the same space.
… I partitioned out the spacial info into separate files. Originally going to use RDF HDT in separate files.
… But some pieces were missing. Did something similar: Apache jena graph, backed by apache arrow.
… RDF mapped to chunks. Predicates being loaded separately.
… That allowed performant viewing.
… There all in their separate worlds.
… Could show you in a couple of weeks. Was using W3C opena nnoatations. But recently changes to geoSPARQL.
… Creating feature collections and specifying other data, heatmaps and segmentation.
… published a paper in 2019, 7k images of several cancer types.
… Did a preload to see how it will work.

allen: https://www.netestate.de/dicom/DICOM_metadata_as_RDF.pdf

erich: preprint: https://arxiv.org/abs/2304.10612

Fixing RDF examples

tim: Jim and I have PRs open that fix a lot of example. I guess I need to post in zulip.

eric: Recently working on something that ports between FHIR JSON and FHIR Turtle, to be able to compare the graph and see if they are the same.

dbooth; Would that be usable in the RDF playground for validation?

eric: No, different code base. Using it for AMIA tutorial.
… Wanted to test against official examples and see where we differ.
… Goal in playground was to not be tied to FHIR resources, to allow us of other logicla models.
… Current work depends on standard FHIR core.

https://github.com/w3c/hcls-fhir-rdf/labels/Turtle

jim: Want to be able to gen one FHIR RDF example at a time

ACTION: tim to make github issue about gen one FHIR RDF exmape at a time

Elment summary

w3c/hcls-fhir-rdf#133

tim: Need a way to connect a Patient class to the StructureDefinition for Patient, for example.
… Suggestion was to put in the class definition in the FHIR RDF ontollgy.

tim: IN RDF we have the distinction bewteen the class and the StructureDef. Need to tie them together.
… Then I need to know which properties are included in the resource summary. That info is in the StructureDef but not in the class.
… That link would them together.

eric: Sounds like the right way to do it.

dbooth: What exactly would the URL be?

eric: We'd need to work on the RDF-ization of the StructreDefs. Algorithm might work out of the box, but not guaranteed.
… No sure they'd be dereferenceable, but we don't control that.

eric: Big prob w examples is that an overhaul to make them resolvable would give you a FHIR server. That would be a very helpful.
… .Would be a great opportunity to enable follow-your-nose.

ACTION: Eric to make github issue about making FHIR examples dereferenceable

tim: This looks like the right URL https://www.hl7.org/fhir/patient.profile.ttl.html So this could go in the Patient class…. Or we could put the Patient class IRI in the profile

tim: Maybe have prov:isDefinedBy https://www.hl7.org/fhir/patient.profile.ttl.html

tim; That's an annotation property.

ACTION: Tim to show example of linking class to structure def, and also show how an objectProperty could be used instead

ADOURNED

Meeting minutes

Concept IRIs

gaurav: Working on this ticket: https://jira.hl7.org/browse/UP-464
… Need to email committee leads to get a vote scheduled.
… We made a change to the FHIR spec for version 5. Added a dash to the names in the naming system fields, but didn't in the code system fields.

https://build.fhir.org/codesystem-namingsystem-identifier-type.html

https://terminology.hl7.org/CodeSystem-v2-0203.html

gaurav: Fix this? We could add a hyphen into the code system. Or if we want to standardize to no hyphen, that would require a change to FHIR.

rob: v2 code system follows that pattern. Why do we need to update a v2 code system?

gaurav: I think there's a LOINC code system.

rob: To match the style of that code system, then without hyphen seems okay.
… It matches the typical convention of that.
… None of the others have a hyphen, so I could see removing the hyphen.
… Codes often use hyphens.
… I'd leave it alone.

dbooth: Benefit in having them both have the hyphen or both without the hyphen.

gaurav: Removing the dash is a change to FHIR, adding the change would be a change to THO.
… Easy thing would be to add the hyphen to THO.

rob: That would be the worst thing.
… Another precedent, style convention, in V2 and V3 codes never have dashes and always upper case, but in FHIR they may have them and may have different case.
… Dashes are common in FHIR.

erich: Agree w consistency.
… Would opt for consistency.

rob: naming system must be lower case. FOr code system, must have upper case.

jim: I like consistency. Sounds like most aligned w convention would be to remove the dash, but don't know how much turmoil that causes.

jim: Changing FIHR spec would cause confusion.

dbooth; Could treat it as an erratum until R6.

rob: Bug fixes to R5 are likely to be scoped to critical changes.

dbooth: How are people leaning?

gaurav: Don't care.

jim: Leaning to leaving it along.

tim: neutral.

rob: on the fence. okay w change proposal.

erich: leave it alone.

dbooth: Anyone opposed to leaving it as is.

AGREED: Leave it as is.

gaurav: Next issue: want to create a PR for next concept IRIs, i think MESH and something else.

Fixes to Turtle examples.

jim: I'll nudge them in zulip. Also made an issue on unit tests.

w3c/hcls-fhir-rdf#136

dbooth: I thought that there were round trip tests, but they got disabled in the build because they took too long.

tim: In R5 we switched to a different parser/generator. The tests are duplicated across FHIR version.

jim: Would be a waste of time to do a custom FHIR RDF parser, given that we have Jena.

dbooth: Need to check w EricP about HAPI FHIR.

tim: Could add some turtle unit test for R5.

dbooth: How many?

tim: WOuld be nice to do the ones that currently have bugs
… I can take this on.

HAPI server

tim: The app I showed you I hooked to a FHIR server, and the HAPI fhir server can give RDF, but it needs to be updated to give R5.
… It says it is giving R5, but it's actually giving R4.

ADJOURNED

Meeting minutes

Drug ID ontology

DBooth: Will follow up with Heiner about scheduling this

Turtle examples

tim: Sent PR and mentioned it zulip.

eric: That will allow us to test shex better.

ACTION: tim to ping grahame on zulip

tim: the other changes i have add things that make it easier to gen all the exapmles without genning the entire FHIR spec.

tim: "I made some progress towards adding extra unit tests for generating examples tmprd/org.hl7.fhir.core#1 "

eric: Could put them on github so I can grab them.
… I'll give you permission to put in in fhircat repo

http://github.com/fhircat/ShExValidation

dbooth: Will these changes fix all known issues in the Turtle examples?

tim: Fixes missing datatype annotations, duplication of root properties. I loaded a lot of these w protege and it told me a lot were not correct. Also there was an issue w URLs having bad chars.
… Some have a vertical bar in them.

tim: Current fixes lined up HL7/kindling#148
… w3c/hcls-fhir-rdf#120

https://chat.fhir.org/#narrow/stream/179177-conformance/topic/syntax.20of.20Reference.2Ereference.20with.20version

eric: Does FHIR protocal accept accepts the percent-encoded vertical bar in a request?

rob; Depends on how the operation is defined.
… If' it's looking for a URI you have to strip it out.

eric: canonical: http://example.org/fhir/questionnaire/patient-intake|1

encoded: http://example.org/fhir/questionnaire/patient-intake%7C1

REST API: http://example.org/fhir/questionnaire/patient-intake?version=1

ACTION: Eric to follow up on the zulip thread

Imaging metadata

erich: Also similar to features pulled from deep learning pipelines. How far does FHIR go to having vocab to describe them?
… Geo groups say it doesn't need to be limited to geo.
… Dicom has a lot of vocab for radiology images, and now has pathol vocab. Recently added annotation capabilities. But there's no official RDF casting to it.
… Which group handles this, and how do i connect it to FHIR RDF data?

rob: in RDF, no clear answer. Between imaging WG and FHIR infra and this group, we'll need to figure it out. Not sure anybody's explored it.

erich: Can do demo in a few weeks.

tim: Have you seen the imaging study resource, and how it connects to DICOM?

erich: Not yet.

https://www.hl7.org/fhir/imagingstudy.html

erich: Some casting from DICOM to RDF have been done, but nothing official. Seems like an opportunity.

dbooth: identified others with similar interest?

erich: OMOP group is interested, but again no official RDF casting.

eric: Where do you want to go?

erich: Based on problems I need to solve.
… Want to base it on standard where possible.

eric: if you ask how to connect to external resources, the FHIR folks will say "let's make FHIR resources for it"
… There's bound to be a boundary on that. Better to have DICOM on the inside or the outside of that boundary?

eric: There will be a swat4ls conference in a couple months. Scott Marshall will be there. Maybe get together there? Maybe push the DICOM people to annoint?

eric: Seems like we need an annointed DICOM RDF

ACTION: EricP to ping Scott Marshall about DICOM RDF

tim: Laterality refers to DICOM. Maybe make an extension of the ImagingStudy resource?

eric: If we already know how to connect the two ontologies, then in theory we can do what we need. FHIR Extensions are not very friendly to RDF use.

ADJOURNED

Meeting minutes

Concept IRIs

gaurav: Emailed ITS committee about minor changes to make to UTG records for LOINC.
… Want to link directly to fhir.org
… But it needs to be approved by ITS. Meeting Wed Jan 3, 3p Eastern. I'll join.
… I'll be working next week, then out the following week. But want to get the MeSH proposal ready for them to vote on that also.
… Also want to pursue radlex, but need to check with them first.

dbooth; Contact for radlex?

rob: Look on the HTA page on confluence. They would have talked with radlex.

https://confluence.hl7.org/display/TA/RadLex

gaurav: URLs in their ontology are resolvable.

Issue 120 canonical URIs.

w3c/hcls-fhir-rdf#120

ericp: Difference bewteen options 1 and 4 is that 1 is intentional and 4 is specific.
… 4 uses proposal B.

tim: This example uses fhir:instantiatesCanonical:

https://build.fhir.org/procedure-example-f201-tpf.ttl.html

rob: Must have dealt with polymorphisma on observationValue

https://build.fhir.org/observation#a14.

eric: 3 ways we could do this. 1 if it's a complex type. 2. It's a primitive type with an XSD datatype, like xsd:datetime.
… 3. It's a primitive, but the xsd is a string and we need a type arc on the blank node.
… valueDateTime is already clear becuase the xsd type is in the value.

tim: Currently the fhir:Reference values of fhir:subject in the examples do not have fhir:link , so that would need to be added, right?

https://build.fhir.org/procedure-example-f201-tpf.ttl.html

ericp: Yes, all the references shold have links.

… ( omplex type)

ericp: Another complex value example: Observation.valueSampleData: https://build.fhir.org/observation-example-sample-data.ttl.html
… This one is a primitive with no xsd type indicated: Observation.valueString: https://build.fhir.org/observation-example-eye-color.ttl.html
… (defaults to xsd:string in turtle)

dbooth: Need to collect these examples, itemize them, then say what we think should happen.

AGREED: Everything that points to something else in FHIR should have a link

ericp: I think that's only Reference and canonical

tim: Do we want to do the version thing in any of the links?

eric: Proposal 4b involves cutting off the version at the pipe, and putting it into a query string.

tim: Should we change those xsd:anyURI types to xsd:string?

ericp: I don't think we should try to be holier than the XML folks.

eric: New issue is how we discriminate polymorphic primitive types

rob: There's a slight different between a choice and value[x] polymorphism.

ericp: I'm guessing that, other than value[x] cases, the only places where it is polymorphic are using Reference or Canonical.

ADJOURNED

Meeting minutes

Introductions

Darrell Woelk: involved in semantic world, going back to MCC. Then worked on Carno, using Cyc. The started Info Sloop, Made a company in 1995, then got into healthcare, building a soft dev platform in healthcare Started w FHIR in 2015. Green Room Tech is our latest. I do FHIR consulting, both tech and business. When Chat GTP cam out, I started working on AI again. Ivolved w U Texax Information, doing lectures on FHIR, then last hyear on gene

rative AI. Had 4 gra dstudents working on gen AI, and want to do more this year.

Darrell's ideas on student involvement.

Darrell's slides: https://lists.w3.org/Archives/Public/www-archive/2024Jan/att-0001/Human_Phenotype_Ontology_and_FHIR.pdf

darrell: Working w Dr Ding.
… KG and LLM.
… Spoke w Juan Sequeda in Data.world. They did soft dev and pub an article about KG and LLM bencharks.
… It can gen SQL, using SQL DL.
… Doing the same thing w an ont and SPARQL
… Juan did a mapping into DB
… They had a claims ont.
… data.world executes the SPARQL queries.
… They turn those SPARQL queries into SQL and return answers.
… They did a benchmark, comparing SQL-only-based vs ont-informed. It was 50% better, and they showed why.
… They used zero-shot prompt.
… They made this available for others to try.
… Proposing to do this in the healthcare domain. On the left side, that meansw we don't have to tell chatGPT what the FHIR resources are, becuase it already knows that.
… Use a FHIR API earch instead of SQL.
… On the RHS want to do something similar to what they did w insurance claim, but w phenotype ont.
… Then using FHIR RDF data.
… Or could export all my FHIR RDF data into the triplestore.

eric: Sounds fun. There might be some code that you could reuse, that does part of that.
… SPARQL query gets dismantled to match hierarchies agains a library of FHIR path. Rule head would be FHIR Path, body would be SPARQL.
… Divide up resources and test it against know FHIR paths. Launch , get results back, bind appropriate resources.

(Eric explains the code more)

eric: That's in typescript right now, Claude Nanjo and I worked on it.

eric: There's a shex schema for FHIR.

dbooth: And a FHIR ont.

darrell: Phenopackets include a lot of data. Vulcan project trying to come up w FHIR rep of phenopacket.
… Met w them yuesterday.

jim: HPO folks have a transformation to FHIR ont.
… They're at least workin on it.
… Someone who works for Chris Chute working on it.

darrell: These are grad students. Class is AI and health. Want a group of 2-3 students. How to intro FHIR RDF?

https://yosemiteproject.org/tutorial-fhir-rdf-as-a-bridge-to-the-semantic-web-in-healthcare/

dbooth: That hasn't been updated to R5 yet.

eric: Did the HPI queries assume that the closures implied by the ont were already instantiated in the SQL data?

darrell: IDK.

eric: could either assume no inference, or could add subsumptions in tables, and join those also.
… Or you could transform the query to add the subsumptions.

darrell: They product is a catalog. Guessing that they're not pushing on inference, but I'm interested in it.
… This would be a first step toward building a test system.

jim: I think the work by Joe Flack on ontology-to-FHIR has been folded into OAK python (Ontology Access Kit): https://incatools.github.io/ontology-access-kit/howtos/fhir-conversions.html

eric: Queries were supposed to go over their info mdel.
… You could infer the structure from the DDL, or maybe from an ont.
… But sounds like they used both. How did they use the ont?

darrell: Their product maps from KG to individual DBs.

eric: Could ask a FHIR query, either 1. "I'm looking for sombody w a procedure performed on the kidney", and that's turned into a query for procedure performed on a kidney, or on kidney lining, etc., using the closure over the subclasses to expand the query.
… 2. Issue the original SPARQL query, but have the infra on the FHIR side, HAPI FhIR server goes to the term server to do that closure.

darrell: And HPO work is doing that?

jim: Joe Flack is the developer working on that at HPO. He works for Chris Chute at Johns Hopkins, but also working on Monarch initiative.
… The OAK python toolkit does lots of things w ont. IDK if they're hosting the HPO ont someplace as FHIR.

eric: Wonder if you could use shex to steer GPT4.

gaurav: You could also give it a shex schema and tell GPT4 to "using this schema, give me ... "

eric: But a well-informed person still needs to read the GPT4 responses to tell it to correct them.. Maybe use shex for that.

eric: Will be hard to make it guess a good query.

darrell: Benchmarks that data.world did are not perfect, an dthey discuss them.
… How to prompt it to correct the mistakes?

eric: What's "included" in the FHIR queries?

darrell: it joins other info to your results.

Concept IRIs

gaurav: The submission part of jira is down right now. Cannot edit it right now.
… But I'll keep an eye on it.

ADJOURNED

Meeting minutes

Introductions

Tsui: Cornell student, in health informatics, data sci, in class of health interoperability. Prof asked students to engage in HL7 WGs, and how they work and importance of standards. Woked 10 years as IT consultant at HP and Oracle, as engineer. AI and ML is growing, want to keep up on latest tech and concepts.

IRI stems

gaurav: UP 464 was stuck because jira crashed. That's now fixed, and this proposal has now been officially submitted.

https://jira.hl7.org/browse/UP-464

gaurav: Next proposal is to add IRI stems to MeSH. That proposal is ready to go. Should review it. Need ITS to vote on it.

https://jira.hl7.org/browse/UP-377

gaurav: After ITS has approved it, I can submit it.

gaurav: MeSH is establishing the template for a single IRI stem. In the future we might propose multiple at once.
… It's linked from our doc page as an example of how to do it.
… Planning next ones including radlex: w3c/hcls-fhir-rdf#123
… Radlex has a standard prefix, but haven't been able to find authoritative info on their website.
… Then Human phenotype ont, and that's an OBO ont.
… Then we might be able to do all OBO onts at once.

ShEx

eric: Improved performance and found input data errors. Will roll that out soon.

DICOM

erich: One of the people working on their version of DICOM-to-RDF. Not open source, but they periodically release the ont.

U Mich work

allen: Making progress. Will be ready to show demo using FHIR RDF in a few weeks.

Next week's HL7 meeting

dbooth: Updating slides for Tuesday's ITS call: https://docs.google.com/presentation/d/1grnUDYgp0QkBsgoNsvvt9v7OUhQ7yRh2OIl6ljbpsks/edit#slide=id.g27e07705fd1_0_0

ADJOURNED

Table of Contents

Attendees

Meeting minutes

ElementDefinition summary flags #133

HL7 working group mtgs

Playground

Summary of action items

Diagnostics

Attendees

Meeting minutes

Update on FHIR RDF

Corrected reverted R5 decisions

Approval of minutes

Diagnostics

- DRAFT -

FHIR RDF

05 Oct 2023

Attendees

Contents

AMIA Tutorial

HL7 Jira

HL7 mtg report

Turtle parser in FHIR Kindling

Summary of Action Items

Attendees

Meeting minutes

Turtle parser/serializer

Concept IRIs

Jira issues

CURIEs

Summary of action items

Diagnostics

Attendees

Meeting minutes

FHIR Turtle parser

Element definition summary flags

Eric's linked data browser

Summary of action items

Diagnostics

Attendees

Meeting minutes

Demo of FHIR RDF app by Tim Prudhomme

ShEx for driving UI

Diagnostics

Attendees

Meeting minutes

Introductions

Relating imaging data to clinical info

Fixing RDF examples

Elment summary

Summary of action items

Diagnostics

Attendees

Meeting minutes

Concept IRIs

Fixes to Turtle examples.

HAPI server

Diagnostics

Attendees

Meeting minutes

Drug ID ontology

Turtle examples

Imaging metadata

Summary of action items

Diagnostics

Attendees

Meeting minutes

Concept IRIs

Issue 120 canonical URIs.

Diagnostics

Attendees

Meeting minutes

Introductions

Darrell's ideas on student involvement.

Concept IRIs

Diagnostics

Attendees

Meeting minutes

Introductions

IRI stems