Stream: omop + fhir
Topic: Updates on FHIR + OMOP collaboration
Ward Weistra (Jun 14 2021 at 11:11):
Hi @Wayne Kubick, is there anything more to share about collaboration between the OHDSI and HL7 community?
Ward Weistra (Jun 14 2021 at 11:13):
Just to make those who I know to be interested aware of this channel:
@Jose Costa Teixeira @Jean Duteau @Josh Mandel @ Bob Milius @Kees van Bochove @May Terry
Kees van Bochove (Jun 14 2021 at 12:12):
Ward Weistra said:
Hi Wayne Kubick, is there anything more to share about collaboration between the OHDSI and HL7 community?
Thanks for raising that @Ward Weistra, I just wrote about this too: https://www.thehyve.nl/articles/omop-fhir-connectathon-reflections. I think a possible next step is to set up an Implementation Guide for mapping FHIR data to OMOP, but nothing set in stone yet.
Ward Weistra (Jun 14 2021 at 12:13):
And from our FHIR DevDays data models discussion, adding @Tia @Vivian Neilley @Sebastiaan van Sandijk
Sebastiaan van Sandijk (Jun 14 2021 at 12:20):
I guess a first step is to decide on the use case(s) we want to work on in the collaboration. There are some possible choices, but I do not know what HL7 and OHDSI have been discussing... https://doi.org/10.5281/zenodo.4474372 gives some options.
Sebastiaan van Sandijk (Jun 14 2021 at 12:21):
Similar usecases were suggested last week in the EHR WG by Paul Nagy
May Terry (Jun 14 2021 at 14:13):
Thanks for including me. Happy to be more involved in this effort. Just for historical context, there was a rather thin IG back in 2018 to map from FHIR to OMOP through the Common Data Model Harmonization (CDMH) project. https://build.fhir.org/ig/HL7/cdmh/profiles.html#omop-to-fhir-mappings. Unfortunately, it stopped in 2018 and is now outdated.
Kees van Bochove said:
Ward Weistra said:
Hi Wayne Kubick, is there anything more to share about collaboration between the OHDSI and HL7 community?
Thanks for raising that Ward Weistra, I just wrote about this too: https://www.thehyve.nl/articles/omop-fhir-connectathon-reflections. I think a possible next step is to set up an Implementation Guide for mapping FHIR data to OMOP, but nothing set in stone yet.
Ward Weistra (Jun 14 2021 at 14:18):
Great to know that background on the Common Data Models Harmonization Implementation Guide, @May Terry . I'll add that to this thread on an overview of all the mappings.
Wayne Kubick (Jun 14 2021 at 14:38):
We skipped out last HL7-OHDSI meeting due to the WGM and DevDays so nothing new to report. But should be able to post an update later this month. Am working to finalize publish the FHIR Common Data Model Harmonization IG which contains a set of OMOP-FHIR mappings.
Ward Weistra (Jun 14 2021 at 14:40):
Thanks @Wayne Kubick, perhaps a caution that May just reported that these mappings are likely outdated. But it can be a good start nonetheless if that is communicated as such.
Grahame Grieve (Jun 14 2021 at 16:53):
what makes them outdated?
Ward Weistra (Jun 15 2021 at 14:52):
@May Terry :point_up: ?
Someone pointed out at least the curious mapping between OMOP's Person.care_site_id (The Care Site refers to where the Provider typically provides the primary care.) to FHIR's BodySite.patient (Reference(Patient), Who this is about on a resource for 'anatomical location of a specimen or body part'). Even separate from the fact that CDMH is an R4 IG and BodySite having been renamed to BodyStructure in R4.
May Terry (Jun 15 2021 at 15:09):
Correction...apologies. I had this flipped. OMOP v5.3.1 does include the DEATH table. OMOP v6.0 removed it.
On the OMOP side, the CDMH OMOP mapping (http://build.fhir.org/ig/HL7/cdmh/profiles.html#omop-to-fhir-mappings) is showing the DEATH table, which implies that it's OMOP v6.0. This version is actually not the preferred version. The next OMOP CDM version is actually branching off of CDM 5.3.1 which does not include or plans to include the DEATH table.
Grahame Grieve said:
what makes them outdated?
May Terry (Jun 15 2021 at 15:56):
The one gotcha overall that I struggle with OMOP CDM and FHIR is on the alignment of controlled vocabularies and value sets. The CDMH mappings were very cursory at the profile level. There is no specific mapping of which exact elements apply, and there is no cohesive strategy on how to handle relevant HL7-specific value sets (e.g.: Observation.category) which are not included in the OMOP vocabulary tables unless you put them all under the source_value field in the respective OMOP table. So, CDMH was a good start for OMOP but it was really thin and not thorough enough for an implementer to use. On a side note, I really enjoyed the AMIA COVID-19 research paper which touched on the integration of FHIR and OMOP CDMs (https://academic.oup.com/jamia/advance-article/doi/10.1093/jamia/ocab108/6276433), but this also didn't provide enough information. Perhaps we can tap @James Agnew to share more about their learnings. I didn't attend FHIR DevDays this time so I don't know what was shared.
Sebastiaan van Sandijk (Jun 15 2021 at 20:14):
About the conversions/mappings being 'outdated' , I think most if not all of these conversions/mappings are rather a work-in-progress. It does not help that indeed OHDSI has recently decided to skip or at least postpone adoption of CDM v6. Apart from that, most mapping/conversion projects that I know still have some clinically relevant Resources or mappings to OMOP that need work. There are also (fixed) mapping choices in several projects that make them less generic in their application.
Sebastiaan van Sandijk (Jun 15 2021 at 20:19):
For me, in the preparation of the FHIR DevDays, this led me to the observation - please challenge me - that there may not be a generic FHIR to OMOP mapping. Such conversions always need a two-step approach, where FHIR is never the actual source of the data, but rather an intermediaire step. This implies that there is always (a) a first local, idiosyncratic, specific conversion from the source to FHIR. Then (b) a second, more generic conversion can be implemented from FHIR to OMOP.
Sebastiaan van Sandijk (Jun 15 2021 at 20:24):
For (a) I think many existing and available source to FHIR conversions can be re-used. For (b) it would be good to make better use of the various opportunities in the FHIR spec for capturing provenance and source values. This is part of OMOP best practices, but has not been given much attention - to my knowledge - in the existing FHIR-omop projects
Sebastiaan van Sandijk (Jun 15 2021 at 20:26):
I am convinced that this will also help to implement more robust bi-directional conversion pipelines.
Sebastiaan van Sandijk (Jun 15 2021 at 20:32):
To be precise:in the above posts I am talking about the FHIR to OMOP ETL usecase. This is only one of the various possible use cases. For example, OHDSI should be able to convert its prediction study results into CDS Hooks cards, and for the patient count use case one could think of OHDSI cohort definitions to be invoked by a $cohort-datarequirements operation?
Grahame Grieve (Jun 15 2021 at 21:56):
thanks @May Terry for the good information about the issues in CDMH. Should we start collecting these issues in a list somewhere?
May Terry (Jun 15 2021 at 22:01):
@Grahame Grieve - that's a good idea. Perhaps HL7 Confluence?
I don't know the whole background or the ask behind the OHDSI/HL7 collaboration other than the announcement but I'm a big proponent on a better alignment between both the OMOP CDM and FHIR models. I am working on COVID19 related projects and a health platform which can accommodate multiple research models so the CDMH work or similar efforts are of great interest to me so happy to help either start it or add to/support what others will start.
Grahame Grieve (Jun 15 2021 at 22:08):
well, creating a confluence page is a great place to start. I think we agreed that we're going to use HL7 confluence? Or was it OHSDI? @Sebastiaan van Sandijk ?
Alexander de Leon (Jun 16 2021 at 01:45):
We are doing some work in this space (OHDSI) at KP, especially around unifying the Common Data Model to our data science work. A colleague is working on this and I am “on assist”, as FHIR is the method of transport. Please include me in this thread.
Sebastiaan van Sandijk (Jun 16 2021 at 20:47):
I do not know @Grahame Grieve I have not been involved with that. I am reaching out in OHDSI to find out who is. Would be good to make a start.
May Terry (Jun 25 2021 at 14:32):
@Christian Reich can best address this for OHDSI.
Sebastiaan van Sandijk said:
I do not know Grahame Grieve I have not been involved with that. I am reaching out in OHDSI to find out who is. Would be good to make a start.
Sebastiaan van Sandijk (Jun 26 2021 at 18:35):
Yes, I contacted him. He said he would send some emails right away....
May Terry (Jul 15 2021 at 15:44):
just checking in since the last post. has there been any traction on the FHIR + OMOP CDM collaboration and alignment? If so, let me know. Would be great to participate and help where I can. Thanks!
Paul Denning (Jul 15 2021 at 18:49):
https://confluence.hl7.org/display/OOF
Great track at NCQA Digital Quality Summit (DQS) about OMOP and FHIR.
https://data.ohdsi.org/PhenotypeLibrary/
https://www.ohdsi.org/software-tools/
Wayne Kubick (Jul 27 2021 at 21:46):
Everyone,
It's finally here! Please join us for a one-hour inaugural open HL7-OHDSI meeting, which will happen on Wednesday, 4th of August 2021, at 10am Eastern Time (16.00 European Time). We want to review (and update) our plans for the collaboration and discuss how you can get involved. Agenda and details will be posted on HL7.org (I'll send a link here when available).
Register at https://hl7-org.zoom.us/meeting/register/tJ0lfumgrTIuGtLuqlYgb4RXXU66J1tlRbf1
After registering, you will receive a confirmation email containing information about joining the meeting.
Wayne Kubick (Aug 02 2021 at 13:56):
Just a friendly reminder about Wednesday's OMOP+FHIR meeting. Additional details are available on the HL7 website: https://hl7-org.zoom.us/meeting/register/tJ0lfumgrTIuGtLuqlYgb4RXXU66J1tlRbf1
Vivian Neilley (Aug 04 2021 at 14:03):
OMOP and HL7 update streaming now! https://hl7-org.zoom.us/meeting/register/tJ0lfumgrTIuGtLuqlYgb4RXXU66J1tlRbf1
Will Rosenfeld (Aug 11 2021 at 18:10):
FYI -
- August 4 Kickoff Meeting: Recording and materials were posted https://confluence.hl7.org/display/OOF/Inaugural+HL7-OHDSI+meeting+on+FHIR+and+OMOP+-+4-August+2021
- End of August Work Group Meeting: Planning 2 hour workshop
For details, see PDF of HL7's email: FHIR-and-OMOP-Webinar-Follow-Up-Email-2021-08-11.pdf
Will Rosenfeld (Sep 13 2021 at 17:49):
FYI - There is a HL7 FHIR OHDSI OMOP event posted for September 29 and 30th on HL7 (http://www.hl7.org/events/)
Dana M Stephenson (Sep 24 2021 at 10:08):
Hi Everyone! For the upcoming HL7 FHIR and OMOP workshop(http://www.hl7.org/events/), we have some great use cases/profiles that will be presented during breakout sessions and are seeking a lead for predictive models and clinical decision support. Please contact me if interested.
Mareike Przysucha (Sep 29 2021 at 11:05):
Hi everyone. I was told today, that I could attend the meeting. Is it still possible to register?
May Terry (Sep 29 2021 at 14:24):
A dedicated chat stream for the OMOP Oncology use case is available here: https://chat.fhir.org/#narrow/stream/302239-omop-.2B.20fhir.20oncology
May Terry (Sep 29 2021 at 16:08):
Thanks all who joined in today's FHIR-OMOP Oncology use case kick-off meeting. For anyone who could not make those meetings and are interested in contributing, reach out and come aboard anytime - look here regularly for updates. We'll be updating our newly established homepage https://confluence.hl7.org/display/OOF/FHIR-OMOP+Oncology+Use+Case
Please feel free to join our initiative as we work towards the goal of aligning FHIR and the OMOP CDM for the purposes of building an oncology learning health system that exchanges patient data for large scale observational studies and analytics.
Dana M Stephenson (Oct 05 2021 at 17:09):
Thanks to all those who participated in last week's workshop. I would like to echo May's inviting words! If you were unable to join, please come aboard anytime. You can access the recordings, materials and resources on the HL7 Confluence page.
We established five (5) subgroups to advance the use cases discussed during the workshop. More could be added as we identify more uses cases over time. You can find subgroup updates and resources on the Projects/Use cases page.
Please feel free to bookmark the FHIR and OMOP Working Group and subgroups homepage to stay up-to-date on all activities and information.
May Terry (Oct 20 2021 at 13:15):
Quick status on the FHIR-to-OMOP Oncology Use Case:
As a follow-up to our organizational knowledge sharing, on Thursday, 10/28 at 1p ET, will be presenting on current initiatives in OMOP Oncology as well as additional discussions on mCODE alignment with OMOP. This will be helpful for the HL7 FHIR-centric side of our team.
The Microsoft Teams collaboration link is below. Hope to see you all there!
https://teams.microsoft.com/l/meetup-join/19%3a239caede18574bc183f75a7a5eb9e826%40thread.tacv2/1608665830064?context=%7b%22Tid%22%3a%22a30f0094-9120-4aab-ba4c-e5509023b2d5%22%2c%22Oid%22%3a%22e1bfdbcb-c8d0-4744-bdf5-a26789ddd15b%22%7d
Jake Gillberg (Oct 21 2021 at 15:20):
Hello folks, I help organize the OHDSI Health Equity workgroup and participate in the OHDSI Vocabulary workgroup. For our next health equity workgroup meeting (Tuesday, November 2nd at 2PM Eastern), we plan to discuss terminologies and mappings for patient-level social determinants of health, risk factors, and needs, and other domains useful for health equity research. I'm interested in seeing how the OMOP+FHIR collaboration could look in this space, especially given the work of project Gravity. Is there anyone here familiar enough with Gravity that would be able to participate in this meeting? Possibly giving a 5-7 minute overview of the work and terminologies?
May Terry (Oct 27 2021 at 23:42):
Quick reminder again for those interested in the FHIR-OMOP Oncology use case to join the OMOP Vocabulary call tomorrow (Thursday, 10/28) at 1p ET for a briefing on the work in OHDSI and OMOP Oncology + with time remaining, a mapping discussion on mCODE and OMOP Oncology CDM. Microsoft Teams link is below.
May Terry said:
Quick status on the FHIR-to-OMOP Oncology Use Case:
As a follow-up to our organizational knowledge sharing, on Thursday, 10/28 at 1p ET, will be presenting on current initiatives in OMOP Oncology as well as additional discussions on mCODE alignment with OMOP. This will be helpful for the HL7 FHIR-centric side of our team.
The Microsoft Teams collaboration link is below. Hope to see you all there!
https://teams.microsoft.com/l/meetup-join/19%3a239caede18574bc183f75a7a5eb9e826%40thread.tacv2/1608665830064?context=%7b%22Tid%22%3a%22a30f0094-9120-4aab-ba4c-e5509023b2d5%22%2c%22Oid%22%3a%22e1bfdbcb-c8d0-4744-bdf5-a26789ddd15b%22%7d
Dana M Stephenson (Dec 06 2021 at 17:29):
Hey everyone! This is a reminder to join us this Wednesday, 8th Dec at 10 ET for the FHIR-OMOP WG All-Hands Meeting. Subgroup leads will provide updates and we will discuss next steps for the WG. Register: https://hl7-org.zoom.us/meeting/register/tJIucuuuqDkoG9VPl3plW09i-Wf2BYMSK_h-
Dana M Stephenson (Dec 13 2021 at 10:14):
Last week's FHIR-OMOP WG all-hands meeting included progress updates from the subgroups and information on upcoming events. You can find the presentation decks and recording on the FHIR-OMOP WG Confluence page: https://confluence.hl7.org/x/njLUB. Please also visit the subgroups (project/use cases) pages for information on meeting times, resources, etc: https://confluence.hl7.org/x/iiq9B
It was also revealed during the call that The Common Data Model Harmonization FHIR Implementation Guide is now available: http://hl7.org/fhir/us/cdmh/
Last updated: Apr 12 2022 at 19:14 UTC
