Building software to convert HL7 v2/v3 messaging into archetypes & templates

Technical (archive)
1

Dear All,

I hope this is the right reflector….

First of all, thanks to those who developed tools, prepared tutorials, etc.!!

I have spending the last few days playing around with the followings:

· Java Reference Implementation of openEHR

· LiU-Archtype-Editor-0.5.2

· Archtype Editor 2.2.779

· ADL 1.5 Workbench beta

· Template Designer 2.6.1213.3

· Etc.

Along with reading very good tutorials on:

· Intro to openEHR, Sam Heard

· Knowledge-enabled approach to eHealth records, Heather Leslie

· Using Archtypes with HL7 Messages and Clinical Documents, Health Frankel

· EN 13606-2 Gello – DCM, Andrew McIntyre

· Etc.

And openEHR stable specifications on:

· Introducing openEHR

· Architecture Overview

· Etc.

And ISO 13606 Part-1 and Part-2.

I truly believe the archtype/template would be the right approach for my project on long-term management and preservation of EHRs. The Java ref. implementation libraries, Archtype Editor, Template Designer are great utils/tools.

My basic question is: are there any public tools available to allow me to covert HL7 v2/v3 messaging to/from archtypes/templates as described in Health Frankel’s tutorial on “Using Archtypes with HL7 Messages and Clinical Documents”?

I know there is deep learning curve and would very much appreciated for any pointers.

Thanks in advance for any help!

–Wo

2

Dear All,

I hope this is the right reflector….

First of all, thanks to those who developed tools, prepared tutorials, etc.!!

I have spending the last few days playing around with the followings:

· Java Reference Implementation of openEHR

· LiU-Archtype-Editor-0.5.2

· Archtype Editor 2.2.779

· ADL 1.5 Workbench beta

· Template Designer 2.6.1213.3

· Etc.

Along with reading very good tutorials on:

· Intro to openEHR, Sam Heard

· Knowledge-enabled approach to eHealth records, Heather Leslie

· Using Archtypes with HL7 Messages and Clinical Documents, Health Frankel

· EN 13606-2 Gello – DCM, Andrew McIntyre

· Etc.

And openEHR stable specifications on:

· Introducing openEHR

· Architecture Overview

· Etc.

And ISO 13606 Part-1 and Part-2.

I truly believe the archtype/template would be the right approach for my project on long-term management and preservation of EHRs. The Java ref. implementation libraries, Archtype Editor, Template Designer are great utils/tools.

My basic question is: are there any public tools available to allow me to covert HL7 v2/v3 messaging to/from archtypes/templates as described in Health Frankel’s tutorial on “Using Archtypes with HL7 Messages and Clinical Documents”?

I know there is deep learning curve and would very much appreciated for any pointers.

Thanks in advance for any help!

–Wo

3

Hi Wo,

Not sure if anyone else has some tools for the Job, but my experience in doing this over the last 5 years is that there is no silver bullet for this task. The process is just like any integration mapping, however the commercial mapping tools that I have attempted to use and those I have attempted to build myself just don’t give me the capability needed for the mapping complexity I require. One of the ley issues is support for abstract types. I would be very happy to know if others have found a better tool than those I have tried

I write XSLT by hand to transform the input message (HL7 V2 needs to be converted to XML using an integration component such as Mirth) into a Template Data Document (TDD) as defined in my presentation. The TDD is validated against the Template Data Schema (TDS) generated from a Template defined in the Ocean Template Designer, which augments the document with fixed and default values in the schema. The validated TDD is then transformed using a TDD to openEHR composition transform that can be used for any template exported using the Ocean Template Designer. These transforms are applied within an integration service which also provides communication ports and message tracing.

You could certainly skip the intermediate TDD step but we have found that using it provides a more concrete (domain) model of the message that you are transforming to and allows the use of standard XML schema validation tools to catch 95% of the structural errors without the need for specialised openEHR validation tools. Although as openEHR experts we benefit from this process, the advantages for those that are not is huge.

The nice thing about using archetypes to define the schema of the target of a transform is that you only need to write the transform template per archetype once. Each message that uses the same input structure corresponding to this archetype can reuse the same transform. Unfortunately, the real world of integration is not so simple and different source systems, even different messages from the same source system may be different, but you still have a good starting point to tweak the transform to support the variation between implementations.

I hope this helps.

Regards

Heath

4

Hello all,

I’m working on my thesis project, which one of the components is EHRstorage in a openEHR repository. Since the input is HL7v2.x messages there isthe necessity of openEHR conversion. I think Heath remembers me, because Heath andChunlan helps me with this issue. I was capable to assemble a small prototypethat includes Mirth, which invokes a Java implementation environment to performthe conversion.

The clinical data that that is present in an Unsolicited ObservationResult (ORU^R01) for instance, it’s successfully converted, but withoutdemographic data, neither a patient identification instance. Can you give someclue about this topic?

Best Regards,
Paulo Ferreira.

5

Hi Paulo,

If I understand correctly, you need a mapping between the HL7 v2.x PID line and some Person archetype of the demographic model: http://www.openehr.org/knowledge/OKM.html#showarchetype_1013.1.479

6

Hello,

Here at the Technical University of Valencia, we have been developing LinkEHR (www.linkehr.com) for the last six years and it seems to be exactly what you are looking for. LinkEHR Studio has two main functionalities:

  • It is a generic archetype editor, able of working with any reference model you import into it. We have worked with archetypes for openEHR, EN13606, HL7 CDA, CDISC ODM, MML and many others.
  • It is a data transformation tool based on archetypes. You can use an archetype and map it to a data source, defining the appropriate transformation functions and rules. Then the tools automatically generates an XQuery to transform existing XML data instances into XML extracts that follow the archetype and reference model rules.

Although we have not worked directly with HL7 v2.x messages, it should be possible to work with them in their XML representation version. You just have to choose or define an archetype as target schema, import the HL7 XML message schema as source, and define the correspondences between those two. We have generated CDA documents based on this methodology, so v2.X messages should be even easier to manage.

Contact us if you have any question.

Best regards,
David

2012/2/3 Chang, Wo L. <wchang@nist.gov>

7

Hi Paulo,
The question is, what are you going to do with the lab report once it is transformed. If you are storing it in an openEHR repository then the is no demographics required in the report. However you do need to find the EHR to store it in, possibly create a demographic record and EHR for the subject. What is necessary is dependent on how you process this workflow. If you use mirth then you don’t need to transform it as you can use the source data unless you need to create a demographic record in which case you perform the same process to do this.
If you are transforming to an EHR extract or similar then you need a wrapper for the TDD with a subject details structure.
We use this later approach for both scenarios because it allows us to define a template data service process independent of any particular integration service.

Heath

8

Dear All: thank you all for your reply to my question! Sorry I was out last week.

Dear Health,

Thank you for taking the time to explain your process getting the HL7 v2/v3 to archtypes & templates!

Would it possible for you to share your XSLT with us? Even better if you can share one or two instances of HL7 messages so that I can follow your instruction steps from your slides to see how the entire process works.

BTW: if anyone knows of any public available test dataset on HL7 v2, v3, 13606, X12, etc. I’d very much appreciated for their URLs.

Much thanks in advanced for your help!

–Wo

9

Dear David,

Thanks for the LinkEHR link especially you have a free version for testing.

I think this is sort of tool that I am looking for to transform HL7, EN13606, X12, etc. into archtypes.

There seems to be a number of LinkEHR video clips available on YouTube, that’s great!

Thanks!

–Wo

10

Dear Paulo,

That’s great that you have your prototype working for the conversion!

If any detail notes and/or codes that you can share, I’d be interested and much thanks in advance for your help.

Best,

–Wo