Stream: analytics on FHIR
Topic: Weekly meetings
nicola (RIO/SS) (Sep 06 2018 at 16:33):
Here is a link to calendar - https://calendar.google.com/calendar/embed?src=health-samurai.io_8eqd81vmm44i6sifb7htmh7hig%40group.calendar.google.com - if you not yet added it
## 6 sep 2018 minutes
- @Ryan Brush published draft SQL-on-FHIR (split into sections for discussion, in future see FHIR tooling for publication - aka fhirpath); move interesting topics from zulip into github issues
- @Patrik Sundberg @Ryan Brush @nicola (RIO/SS) - prototype "right" schema implementation (Spark, BQ, Pg) - demo scenarios?
- @Mikhail Lapshin working on batch loader scenario into PostgreSQL
- @nicola (RIO/SS) - fhirpath in database
- @Bryn Rhodes - collaboration with http://wiki.hl7.org/index.php?title=201809_Clinical_Reasoning (same dataset & measures as queries)
TODO: book room for discussion
May be i forgot something - please add.
@Bryn Rhodes i loose links from hangout :(
Ryan Brush (Sep 06 2018 at 21:43):
For those interested, we're tracking some updates to the SQL on FHIR proposal from the above discussion on GitHub:
- Store precision as a sibling field https://github.com/rbrush/sql-on-fhir/issues/5
- Add reference fields that can be equi-joined: https://github.com/rbrush/sql-on-fhir/issues/6
- Use a native datetime type with a sibling field describing precision: https://github.com/rbrush/sql-on-fhir/issues/7
Ryan Brush (Sep 06 2018 at 21:43):
Also, as a potential scenario for the Connectathon, we discussed attempt to write SQL queries that can implement some of the example measures. Bryn shared these as a reference:
https://github.com/cqframework/cqf-measures/tree/master/pages/cql
And also this test data that could be loaded into SQL tables and used to test the above measures:
http://build.fhir.org/ig/cqframework/hedis-ig/test-data.html
Josh Mandel (Sep 06 2018 at 23:01):
I just want to give a quick shout-out to the awesome communications happening here :-) Even though I haven't been able to engage in this work, it's easy to keep track of what's happening.
Ryan Brush (Sep 06 2018 at 23:28):
Thanks, Josh! It's kind of you to take the time to post this....and I'm glad things are (mostly) making sense. :)
Bryn Rhodes (Sep 10 2018 at 20:47):
Here is the latest test data we use to run the HEDIS measures: https://drive.google.com/open?id=1g24qdYoKkG6BeGMZkTJd7lgbEmLTZOyi
Bryn Rhodes (Sep 10 2018 at 20:48):
We are working on making sure they work with the CMS measures now and will post updates if we end up making any.
nicola (RIO/SS) (Sep 13 2018 at 15:52):
Today meeting was replaced with zulip report. See everybody next week.
nicola (RIO/SS) (Sep 13 2018 at 15:53):
I'm working on fhirpath in database demo.
Mikhail Lapshin (Sep 13 2018 at 15:56):
I still working on resources import from Bulk API server/local files, trying to optimise import speed. Successfully imported Cerner dataset.
Bryn Rhodes (Sep 13 2018 at 16:07):
We have found that we need Encounters more often for the CMS measures than we did for the HEDIS ones, so we are working on adding more Encounters to the test data, specifically ones coded from the CMS value sets.
Ryan Brush (Sep 13 2018 at 17:18):
My updates:
I'm largely working on bringing the Bunsen project (https://github.com/cerner/bunsen) closer to the current SQL on FHIR draft.
For those interested, comments on storing numeric precision (https://github.com/rbrush/sql-on-fhir/issues/5), timestamp precision (https://github.com/rbrush/sql-on-fhir/issues/7) and references that can be equi-joined (https://github.com/rbrush/sql-on-fhir/issues/6) are welcome. We're still very early and will continue revising the approach, but having a working document for these can help prototypes and the connectathon efforts.
Bryn Rhodes (Sep 16 2018 at 21:26):
An updated data set with more Encounter data is available here: https://drive.google.com/file/d/17onxvSMeQxfK5ugHY0-4yzxTSZeQCQzl/view?usp=sharing
Bryn Rhodes (Sep 19 2018 at 00:02):
Found and fixed a few issues with the test data set, update is available here: https://drive.google.com/open?id=1lrBoPyjCOcBkKA_-jhiIDowiwDfFXftt
nicola (RIO/SS) (Sep 20 2018 at 15:57):
Join us in 5 minutes. https://meet.google.com/zvs-ucbe-ufk
Bryn Rhodes (Sep 20 2018 at 16:00):
I won't be able to make the meeting today, I have a conflict. My only update is that we updated the data set, links above. We are close to being able to evaluate the three measures with that data set, and I'm working on a way to translate them to SQL so we can evaluate them in this track.
nicola (RIO/SS) (Sep 20 2018 at 16:22):
@Bryn Rhodes can you share the first measure you are interested in - we will help you
Bryn Rhodes (Sep 20 2018 at 16:25):
We're working on CMS130 right now: https://github.com/cqframework/cqf-measures/blob/master/pages/cql/cms130_FHIR-7.2.000.cql
Bryn Rhodes (Sep 20 2018 at 16:26):
That's a manual translation to FHIR of the current CMS130: https://ecqi.healthit.gov/ecqm/measures/cms130v7
Ryan Brush (Sep 20 2018 at 16:27):
My connectathon-related update: I'd like to get some of the content Bryn has shared running in Bunsen, our open source FHIR-and-Apache Spark SQL project. It doesn't quite match the SQL on FHIR proposal at this point, but it's close enough we can hand edit the SQL for this content.
I mentioned on the call that I'd share references to the Bunsen project, so here they are...
The project documentation is at https://engineering.cerner.com/bunsen
The source code is at: https://github.com/cerner/bunsen
A tutorial to get this running in a Docker container (which I used at the most recent DevDays) is here: https://github.com/cerner/bunsen-tutorial
The Docker container and tutorial is probably the easiest way to get this running if you don't have an environment with Apache Spark 2.3.
nicola (RIO/SS) (Sep 20 2018 at 20:06):
@Mikhail Lapshin did a nice tutorial for bulk load data into postgresql - https://github.com/fhirbase/fhirbase/blob/master/README.md, which covers one of our scenarios
@nicola (RIO/SS) prepared docker container with postgresql and alpha fhirpath extension with easy to use web ui - https://fbdemo.aidbox.app/
and stared post about fhirpath in database - https://aidbox.gitbook.io/fhirpath-pg/
Patrik Sundberg (Sep 20 2018 at 23:32):
Our code is here: https://github.com/google/fhir
The most interesting example to run is still https://github.com/google/fhir/tree/master/examples/bigquery
but i hope to have an example with extensions and profiles ready by next week. fingers crossed.
nicola (RIO/SS) (Oct 08 2018 at 05:42):
@Patrik Sundberg , @Ryan Brush Should we start our status meetings again? May be once in every 2 weeks?
Ryan Brush (Oct 14 2018 at 15:40):
I'd be up for biweekly status meetings on this; we can increase the cadence if necessary when we approach future events. (Also, apologies for the slow response here; I've been out for a bit. )
Yunwei Wang (Oct 15 2018 at 13:56):
Hi, I am new to this stream. What would be the topics for such weekly meeting?
nicola (RIO/SS) (Oct 15 2018 at 16:09):
@Yunwei Wang Read our notes from prev meetings in this thread
nicola (RIO/SS) (Oct 16 2018 at 14:37):
Let's have a meeting next week to updated status?!
Last updated: Apr 12 2022 at 19:14 UTC
