14:55:59 RRSAgent has joined #hcls 14:56:03 logging to https://www.w3.org/2023/11/02-hcls-irc 14:56:06 rrsagent, make logs public 14:56:09 Meeting: FHIR RDF 14:56:12 Chair: David Booth 15:00:39 Present: Avanti Paturkar, Allen Flynn, Erich Bremer, EricP 15:01:14 ACTION: DBooth to put full zoom info on RDF mtg page, for dial-in 15:01:47 Present+ Rob Hausam 15:02:58 Topic: Introductions 15:04:37 Erich: dep of informatics, working on RDF 15 years, working on software for teachers, bridging to clinical . Stonybrook Univ. 15:06:15 Present+ Jim Balhoff 15:06:22 Present+ Gaurav Vaidya 15:09:18 Topic: Relating imaging data to clinical info 15:09:46 erich: Tagging onto OMOP group and, diacon group. Trying to integrate spacial info, and make heatmaps. 15:10:00 ... I'm using RDF, but want to walk it to clinical info also. 15:10:26 ... Any work done w diacom or OMOP? 15:10:42 eric: Some of the same people worked on an OMOP mapping a while ago. 15:11:11 ... Didn't do anything between images and OMOP. Not aware of images to OMOP. Maybe Scott Marshall? 15:11:40 erich: Imaging and clinical domains seem to be separated. Want to bring them together. 15:11:53 ... OMOP is tyring to use Daicom verbiage. 15:12:13 ... Have annotations by the files are usable, but cannot bring up 3M image iunless there's spacial info. 15:12:27 ... Cannot decode the image and put it on the screen quickly. Need spacial info. 15:12:40 ... Same applies to deep learning pipelines. 15:13:00 eric: Write-up of baby use case for clinical setting? 15:13:28 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. 15:13:48 ... Diacom is noce, but unless you set standards of what should be included, you get a lot of empty boxes. 15:14:05 ... Noe of these groups are bridgeing each other. 15:14:41 dbooth: Need to get folks together to duscuss? 15:15:10 erich: David Kume said thatere;s some diacom, but no official mapping. Series of XML docs, you need to extract info from them. 15:15:31 ... FHIR is better, OMOP has CSV which is better. Need to convince people that this is needed. 15:15:47 ... Easier to build something to show them the value, to get them hungry to discuss. 15:15:59 ... Otherwise OMOP group has 100 people. 15:16:21 ... Any image could have 1M segmentations, with lots of info associated. 15:16:33 ... Cannot use standard RDF DB -- billions of triples. 15:16:52 ... Need to break into separate storage, then need to get clinical date into it. 15:17:20 ... In semweb, lots of greag components, but nobody knows what it becomes yet. 15:18:37 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? 15:19:32 erich: Looking for low hanging fruite. Collected a lot of COVID imagery 80-100M radiology images. 15:19:32 ... Wrote software to extract diacom data and RDF-ized it. Loaded into virtuoso to query. 15:19:43 ... Diacom allows linking by accession IDs, dates, etc. 15:20:01 ... Brought into relational data laek, but how to connect them back? Some terms connect, but not all. 15:20:55 ... 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. 15:22:03 dbooth: Maybe show what you have so far? 15:22:45 erich: Could show soon as use case. ONe odball thing: images range from 300M to 3GB, and my computer gets chewed up. 15:23:12 ... Having the pipeline data from deep learning, in central RDF DB is not scalabl, virtuoso w TB of RAm. 15:23:40 ... With a lot of these files, putting into central store, you're doing spacial indexing, but considering it all part of the same space. 15:24:02 ... I partitioned out the spacial info into separate files. Originally going to use RDF HDT in separate files. 15:24:24 ... But some pieces were missing. Did something similar: Apache jena graph, backed by apache arrow. 15:24:39 ... RDF mapped to chunks. Predicates being loaded separately. 15:24:49 ... That allowed performant viewing. 15:25:05 ... There all in their separate worlds. 15:25:28 ... Could show you in a couple of weeks. Was using W3C opena nnoatations. But recently changes to geoSPARQL. 15:25:49 ... Creating feature collections and specifying other data, heatmaps and segmentation. 15:26:07 ... published a paper in 2019, 7k images of several cancer types. 15:26:17 ... Did a preload to see how it will work. 15:27:18 allen: https://www.netestate.de/dicom/DICOM_metadata_as_RDF.pdf 15:27:31 erich: preprint: https://arxiv.org/abs/2304.10612 15:29:00 Topic: Fixing RDF examples 15:29:51 tim: Jim and I have PRs open that fix a lot of example. I guess I need to post in zulip. 15:30:16 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. 15:31:04 dbooth; Would that be usable in the RDF playground for validation? 15:31:23 eric: No, different code base. Using it for AMIA tutorial. 15:31:48 ... Wanted to test against official examples and see where we differ. 15:32:04 ... Goal in playground was to not be tied to FHIR resources, to allow us of other logicla models. 15:32:31 ... Current work depends on standard FHIR core. 15:35:13 https://github.com/w3c/hcls-fhir-rdf/labels/Turtle 15:36:28 jim: Want to be able to gen one FHIR RDF example at a time 15:36:59 ACTION: tim to make github issue about gen one FHIR RDF exmape at a time 15:43:09 Topic: Elment summary 15:43:11 https://github.com/w3c/hcls-fhir-rdf/issues/133 15:43:33 tim: Need a way to connect a Patient class to the StructureDefinition for Patient, for example. 15:44:07 ... Suggestion was to put in the class definition in the FHIR RDF ontollgy. 15:45:08 tim: IN RDF we have the distinction bewteen the class and the StructureDef. Need to tie them together. 15:45:34 ... Then I need to know which properties are included in the resource summary. That info is in the StructureDef but not in the class. 15:45:53 ... That link would them together. 15:46:16 eric: Sounds like the right way to do it. 15:46:40 dbooth: What exactly would the URL be? 15:47:04 eric: We'd need to work on the RDF-ization of the StructreDefs. Algorithm might work out of the box, but not guaranteed. 15:47:27 ... No sure they'd be dereferenceable, but we don't control that. 15:49:04 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. 15:49:24 .. .Would be a great opportunity to enable follow-your-nose. 15:54:29 ACTION: Eric to make github issue about making FHIR examples dereferenceable 15:56:13 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 15:57:46 tim: Maybe have prov:isDefinedBy https://www.hl7.org/fhir/patient.profile.ttl.html 16:01:06 TallTed has joined #hcls 16:01:28 tim; That's an annotation property. 16:03:47 ACTION: Tim to show example of linking class to structure def, and also show how an objectProperty could be used instead 16:03:53 ADOURNED 16:04:37 rrsagent, draft minutes 16:04:38 I have made the request to generate https://www.w3.org/2023/11/02-hcls-minutes.html dbooth