Has anyone attempted to export synthetic data in openEHR format? It’s a treasure trove for anyone testing stuff as well since it’s realistic data possibly answer some preliminary research questions and understand feasibility.
@christian
@bna worked with some students on this a few years ago and it has been talked about before. Perhaps it would be a quick win for FHIR-Connnect. Synthea in FHIR->openEHR
Thanks Ian, yeah I found old threads but nothing tangible. I’ll reach out to involved folk and see if there’s anything reusable.
Also agreed, it’s a winner ingredient for testing and validating any openEHR to/from FHIR transforms. If nothing I think I’ll dive into it myself.
@Koray_Atalag did you get anywhere with this? I have been looking and think I may have an approach to getting sample openEHR data.
Maybe one day it will use FHIR-Connect, but I think that’s still got a way to go before all the wrinkles are sorted out with it.
I’d be really interested in this and need to work on it imminently. I had a wee play with the nhs fork of synthea . @johnmeredith might also have an Interest. The flat format /import call might be useful.
Hi Richard, I’m afraid not. Just didn’t have the time. I honestly think it’ll need to be a team effort if we want something real and useful.
Ah, OK, let’s see how far I get.
I suspect the real challenges will (as ever) be the Observations. I may be back for advice in a few places ![]()
For observation we have a few common questionsloinc to openehr mappings from openrhr Finland that we can share
I’ve worked out how the code works at a high level and have actually started making changes (the toughest bit was getting to grips with Java
)
At the moment, I am anglicising it and aligning it to the UK Core FHIR profiles (as opposed to US Core).
The tricky part (maybe ?) is when I start to emit openEHR compositions. As the data Synthea produces is very episodic, all the data is focused around a series of ‘Encounters’ - For me, I’m going for a mixed mode of FHIR and openEHR, so I will need to emit both FHIR resources and openEHR compositions.
I’m not sure how I can create one composition that contains all the clinical content that Synthea creates. I’m pondering on that, but first I’ll get it aligned to UK Core.
Is it necessary to create one composition for the whole clinical bundle, or could you simplify the problem by creating multiple compositions and wrap them in a contribution?
Sorry if this is a dumb question, I haven’t done a lot of practical modeling myself but I’m interested in the outcome of this project.
If you were emulating a ‘proper’ openEHR source of truth EHR, then you would normally split the data into multiple composition and contribution wrapper, but that does make life more complex.
Part of the work that @heidi.koikkalainen is doing with the IPS template on CKM, is to re-factor it so that individual sections can be reused, either in multiple templates or in a single composition.