FHIR Chat · Sample FHIR data JSONs · implementers

Stream: implementers

Topic: Sample FHIR data JSONs

Paul S (May 22 2019 at 16:44):

Hi Everyone,

Can someone share sample FHIR data JSONs from different vendors? Thanks.

Paul

Michele Mottini (May 22 2019 at 16:48):

I think you can get those from publicly available test servers: http://wiki.hl7.org/index.php?title=Publicly_Available_FHIR_Servers_for_testing

Paul S (May 22 2019 at 16:54):

Thanks @Michele Mottini

Josh Mandel (May 22 2019 at 23:34):

In terms of real world data I can also point to my own at https://github.com/jmandel/my-health-data/tree/master/my-data (so far this is just from Epic systems).

Vadim Peretokin (May 24 2019 at 08:44):

That is cool. Thanks Josh

Vadim Peretokin (May 24 2019 at 08:44):

I hope you don't collect much more, though

James Agnew (May 24 2019 at 12:58):

Are you accepting pull requests?

Clark C. Evans (Jul 06 2020 at 17:52):

This is probably a FAQ.. but is there a large set of sample FHIR JSONs that are suitable for testing eCQM implementations; ideally one that's based upon SynPUF and BlueButton? Yes, I'm aware of Synthea, however, it's not clear what sort of measures that it could support. For example, I ran a 10k patient sample using default settings, and not one of them had a hypertension diagnosis . Thank you in advance for any thoughts/pointers. Has anyone converted DE-SynPUF to FHIR?

Grahame Grieve (Jul 06 2020 at 21:45):

I don't know of anything. @Jason Walonoski @Bryn Rhodes are the most likely to know?

Bryn Rhodes (Jul 06 2020 at 21:56):

These aren't large data sets, but they are tuned to specific eCQMs: https://github.com/DBCG/connectathon/tree/master/fhir3/supplemental-tests

Bryn Rhodes (Jul 06 2020 at 21:57):

And they were generated with Synthea modules for the eCQMs, so you can use them to generate larger sets.

Eric Whitley (Jul 07 2020 at 16:23):

@Bryn Rhodes Sorry to ask this, but I'm struggling to connect the dots. Were there specific existing Synthea modules/rules you were using to generate sample data for the eCQMs?

Clark C. Evans (Jul 07 2020 at 16:28):

@Bryn Rhodes Brilliant! Not only does this have sample FHIR bundles, but a set of eCQMs and expected outputs! Thank you so, so much. Are there any copyright issues I have to be concerned about? I need to repackage this into Julia's Pkg.Resources for further distribution (and use as basis of a regression test). I also want to thank you for CQL (and @Jason Walonoski for his excellent Synthea).

Bryn Rhodes (Jul 07 2020 at 16:33):

You're very welcome, and I'll second the thanks to the Synthea team, they're the ones that produced the modules that support these patient data! @Eric Whitley , if you dig in the Synthea repository, look at the modules folder, it has the module descriptions for a ton of things, among them the eCQM modules that built the data in the connectathon repository.

Bryn Rhodes (Jul 07 2020 at 16:42):

@Clark C. Evans , yes, the eCQM content is creative commons, but use of the value set content requires appropriate licensing

Clark C. Evans (Jul 07 2020 at 16:42):

@Bryn Rhodes These are for STU3; How hard would it be to make a set for R4? Our experiments are so far using R4 (but I could back-port to R3). By creative commons, you mean CC0 (aka public domain)?

Bryn Rhodes (Jul 07 2020 at 16:46):

There are fhir401 here

Bryn Rhodes (Jul 07 2020 at 16:46):

Yes, public domain

Bryn Rhodes (Jul 07 2020 at 16:47):

Note there are fhir4 bundles as well, but the difference between those is really CQL 1.3 (for fhir4) and CQL 1.4 (for fhir401).

Clark C. Evans (Jul 07 2020 at 16:49):

Ok; these don't have the ten numerator / ten denominator Synthea patients though? It seems like one patient per-permutation of the logic branching? Anyway, perhaps I should be coming to next virtual connectathon... looks like you all were having some fun.

Eric Whitley (Jul 07 2020 at 18:20):

@Bryn Rhodes Thank you! This is exactly what I was looking for. Very much appreciated!

Jason Walonoski (Jul 07 2020 at 21:38):

Regarding Synthea:

For example, I ran a 10k patient sample using default settings, and not one of them had a hypertension diagnosis .

Synthea routinely produces about 25-30% hypertension rate. So you might have been looking for the wrong thing.

Jason Walonoski (Jul 07 2020 at 21:41):

It should show up with the SNOMED code 59621000

Jason Walonoski (Jul 07 2020 at 21:44):

Also, I would add this... if there are some "low-hanging fruit" changes we could make that would produce data for more eCQM or CQL testing, we'd be glad to consider them (priority is always an issue, but if I don't know I can't do it). Alternatively, if others develop modules that produce eCQM-suitable data we always take pull requests.

Jason Walonoski (Jul 07 2020 at 21:46):

Although, I may be confusing the issue since there is a Synthea master repo, and the Tacoma fork with lots of eCQM modules.

Clark C. Evans (Jul 08 2020 at 00:17):

@Jason Walonoski Thank you; and of course you are correct. I think as CQL is used everywhere and eCQMs provide lovely example CQL queries; therefore, having data suitable for eCQMs would be very useful, at least for training purposes. That said, I'm not doing any production work with FHIR at the moment, so my perspective may not be common.

Clark C. Evans (Jul 09 2020 at 18:46):

@Jason Walonoski When you have a moment, I have got a Synthea question. How can you specify more than one module to use; I've tried using -m more than once, and it only gives me the last module; I've also specified comma or space separated within quotes without success. When you specify more than one module, how does that work? Thank you in advance.

Jason Walonoski (Jul 09 2020 at 20:52):

@Clark C. Evans I need to update the wiki on this. You need to use the java file system path separator. On UNIX systems, this character is : and on Microsoft Windows systems it is ;.

Jason Walonoski (Jul 09 2020 at 20:55):

So, for example on Mac OS X, using two filters -- one matching "metabolic" and the other "cancer"

./run_synthea -m metabolic*:*cancer*

Results in:

Modules: Breast_cancer Module
       & Colorectal Cancer Module
       & Lung Cancer Module
       & Metabolic Syndrome Disease Progression Module
       & Metabolic Syndrome Standards of Care Module
       & Veteran Lung Cancer Module
       & Veteran Prostate Cancer Module
       > [7 loaded]

Clark C. Evans (Jul 11 2020 at 16:48):

I had a question about DateTime loading from FHIR; and it seems to be addressed at https://cql.hl7.org/09-b-cqlreference.html#todatetime and https://cql.hl7.org/15-h-timeintervalcalculations.html#difference-calculations

Richard Stanley (Feb 09 2021 at 19:20):

Hi @Clark C. Evans I'm curious if you found out if it is possible to load more than one local module at a time.

Last updated: Apr 12 2022 at 19:14 UTC