{Disarmed} Comparison of EHR models

Technical (archive)
1

Dear all,Recently an article by Bernd Blobel was published in the Dutch HL7 magazine (Dec 07 issue) in which he compares the different EHR models: openEHR, HL7v3, EN/ISO 13606 and CCR. Robert Stegwee, the chair of HL7.nl, kindly translates this article in Dutch, which unfortunately makes it unsuitable for distribution outside the Netherlands

I’ve tried to ask Bernd Blobel to share the original text of this article (which is hopefully in English), so that the openEHR community also can take notice of it. I haven’t received an answer yet.

I won’t translate the whole article back to English (and I still hope that Dr. Blobel will share the original article), but for the sake of discussion I would like already to point out a few things that ‘triggered’ me.

From what I understand Blobel claims that all the paradigms for an advanced EHR architecture were, already back in the nineties, defined in the context of the Generic Component Model (GCM) (no reference provided).
In the article he states that the GCM provides a service oriented, model driven system architecture for the development of a sustainable and semantic interoperable EHR systems.
The GCM provides a multi-model approach for EHR architectures, system development and implementations by the simplification of the system description by means of:

  • transparent domain management,
  • the composition and decomposition of the system components
  • the views from the different angles on the system (amongst which thorough modelling of business models

As a result the GCM provides reference architecture for analysing, designing en implementation of EHR architectures, as well as a tool for the development of migration strategies (Educational challenges of health information systems’ interoperability. B. Blobel, Methods Inf Med 2007; 46 p.52-56)

Although I can’t assess the article fully on it’s merits, the idea of a theoretical ‘meta’ reference architecture for the future which can be used for the purposes above seems appealing, both for further improvement of the openEHR architecture as well as for the future harmonisation of HL7 en openEHR in a common (EN/ISO 13606 derived?) internationally accepted EHR standard.

So my first question is: Is the GCM to be seen as a theoretical ‘meta’ reference architecture, which can be used as a guideline for future developments? If the answer is no, why not?

Further in the article Blobel compares GEHR against the GCM. Although the header of this section mentions the openEHR foundation, he consistently talks about GEHR and the GEHR project (MailScanner has detected a possible fraud attempt from “http:www.gehr.org” claiming to be http:www.gehr.org). The URL for GEHR links to a site, which has to do with different aspects of healthcare than we’re generally talking aboutJ). Also when Blobel talks about ADL he refers to a URL that doesn’t exist anymore (http://www.deepthought.com.au/) and most certainly it wouldn’t have linked to the latest version of the ADL

So my second question is: is Blobel, when making his comparison, referring to the latest versions of the reference architecture and ADL as recently developed within the openEHR community?

Blobel’s conclusion of comparing GEHR to the GCM, is that GEHR limits itself to the structural aspects of the knowledge components and doesn’t comprise behavioural aspects. Also it isn’t possible, due to the lack of specified rules, to aggregate archetypes. Instead they have to be replaced by more complex archetypes.
More generally the GEHR approach has some essential shortcomings at the mathematical, system theoretical and informatics levels. These shortcomings have to be addressed in the future.

In the discussion en conclusion section Blobel adds to this: that within the EN/ISO 13606 approach, although almost complete as far as semantic interoperability concerns, a lot of shortcomings and inconsistencies have to be solved. As example: the issue of structural composition and decomposition, as well as the modelling of business processes is not solved well.

Personally I think that such statements should be underpinned with arguments/ scientific proof and/or examples or at least a reference to a properly peer reviewed article that does so. I would like to invite Blobel (and others if they feel obliged to) to provide these scientifically valid facts to underpin these statement, so we can have a proper discussion. This type of ‘review’ statements creates confusion, which hinders any serious discussion about future developments and harmonisation. It also undermines the (in my opinion) otherwise good intention of the article as a whole.

My third and last question to the community is: are these conclusions (if applicable to the current version of openEHR) valid and if yes how can we address those issues?

Cheers,

Stef

2

Hello Stef,

Let me clarify this article: I've looked at the Dutch article you
refer to and being familiar with the publications of Bernd Blobel, I
think this a translation of an English paper he has done on these
comparisons earlier [http://www.ncbi.nlm.nih.gov/pubmed/16964348\] or
maybe an updated version.

The Generic Component Model is designed by Mr Blobel and often
described in his publications, however, not sufficient detailed to
follow his line of reasoning.

The GCM provides a multi-model approach for EHR architectures,
system development and implementations by the simplification of the
system description by means of:
- transparent domain management,
- the composition and decomposition of the system components
- the views from the different angles on the system (amongst which
thorough modelling of business models

This bears references to the general RM-ODP framework. I have not yet
been able to find the differences between GCM and RM-ODP.

Further in the article Blobel compares GEHR against the GCM.
Although the header of this section mentions the openEHR foundation,
he consistently talks about GEHR and the GEHR project. The URL for
GEHR links to a site, which has to do with different aspects of
healthcare than we’re generally talking aboutJ).

The domain name registration must have expired at some point, since it
did point to a website of that project once.

Also when Blobel talks about ADL he refers to a URL that doesn’t
exist anymore (http://www.deepthought.com.au/) and most

Same here. This used to be a website of Thomas Beale.

In the discussion en conclusion section Blobel adds to this: that
within the EN/ISO 13606 approach, although almost complete as far as
semantic interoperability concerns, a lot of shortcomings and
inconsistencies have to be solved. As example: the issue of
structural composition and decomposition, as well as the modelling
of business processes is not solved well.

Personally I think that such statements should be underpinned with
arguments/ scientific proof and/or examples or at least a reference
to a properly peer reviewed article that does so. I would like to
invite Blobel (and others if they feel obliged to) to

This is the general feeling I'm left with when reading an article by
Mr. Blobel.

My third and last question to the community is: are these
conclusions (if applicable to the current version of openEHR) valid
and if yes how can we address those issues?

I think his conclusions are too generic to be able to address them
properly due to lack of sufficient scientific underpinning.

Bye, Helma

3

I have not managed to obtain a copy of the article in question so am going on your summary here and a few other emails I received. Main points:

  • There are two main aspects to building systems: the semantic and the engineering.

  • The semantic aspect is that which enables us to build the first copy of a system that functions as we want. This contains nearly all the hard intellectual design thinking, ontological aspects and domain-related elements. Higher levels of the semantic dimension include business processes.

  • The engineering aspect is about how to turn a single prototype system into a production quality system and deploy it hundreds or thousands of times. Most of Bernd’s work is in this area - the service architecture, security, scalability and so on.

Architectural approaches which only focus on one or other of these aspects won’t produce a widely usable outcome. openEHR has mostly, historically, focussed on the bottom level semantics, and is now focussing on the upper-level semantic aspects and the engineering aspects (mainly service models). See this page for a brief explanation: http://www.openehr.org/201-OE.html

I am not sure why Bernd is saying anything about Gehr - the last time we touched it was at least 5 years ago, and the architectures and understanding we have of solutions for the domain are so radically improved as to make any analysis of Gehr a waste of time. The most useful document for Bernd to read would be the openEHR Architecture Overview - see http://www.openehr.org/releases/1.0.1/html/architecture/overview/Output/overviewTOC.html (PDF → http://www.openehr.org/releases/1.0.1/architecture/overview.pdf)

  • thomas beale

Stef Verlinden wrote:

(attachments)

OceanCsmall.png

4

There is an English version of the comparison. I am sure Bernd would be
happy to share. I think the article has some excellent thoughts and is not
biased toward any one approach. I don't agree with everything, but I found
the article useful.

Ed Hammond

             Thomas Beale
             <thomas.beale@oce
             aninformatics.com To
             > For openEHR technical discussions
             Sent by: <openehr-technical@openehr.org>
             openehr-technical cc
             -bounces@openehr.
             org Subject
                                       Re: {Disarmed} Comparison of EHR
                                       models
             01/19/2008 03:23
             PM
                                                                           
             Please respond to
                For openEHR
                 technical
                discussions
             <openehr-technica
              l@openehr.org>
                                                                           
I have not managed to obtain a copy of the article in question so am going
on your summary here and a few other emails I received. Main points:
      There are two main aspects to building systems: the semantic and the
      engineering.
      The semantic aspect is that which enables us to build the first copy
      of a system that functions as we want. This contains nearly all the
      hard intellectual design thinking, ontological aspects and
      domain-related elements. Higher levels of the semantic dimension
      include business processes.
      The engineering aspect is about how to turn a single prototype system
      into a production quality system and deploy it hundreds or thousands
      of times. Most of Bernd's work is in this area - the service
      architecture, security, scalability and so on.
Architectural approaches which only focus on one or other of these aspects
won't produce a widely usable outcome. openEHR has mostly, historically,
focussed on the bottom level semantics, and is now focussing on the
upper-level semantic aspects and the engineering aspects (mainly service
models). See this page for a brief explanation:
http://www.openehr.org/201-OE.html

I am not sure why Bernd is saying anything about Gehr - the last time we
touched it was at least 5 years ago, and the architectures and
understanding we have of solutions for the domain are so radically improved
as to make any analysis of Gehr a waste of time. The most useful document
for Bernd to read would be the openEHR Architecture Overview - see
http://www.openehr.org/releases/1.0.1
/html/architecture/overview/Output/overviewTOC.html (PDF ->
http://www.openehr.org/releases/1.0.1/architecture/overview.pdf)

- thomas beale

Stef Verlinden wrote:
      Dear all,
      Recently an article by Bernd Blobel was published in the Dutch HL7
      magazine (Dec 07 issue) in which he compares the different EHR
      models: openEHR, HL7v3, EN/ISO 13606 and CCR. Robert Stegwee, the
      chair of HL7.nl, kindly translates this article in Dutch, which
      unfortunately makes it unsuitable for distribution outside the
      Netherlands

      I’ve tried to ask Bernd Blobel to share the original text of this
      article (which is hopefully in English), so that the openEHR
      community also can take notice of it. I haven’t received an answer
      yet.

      I won’t translate the whole article back to English (and I still hope
      that Dr. Blobel will share the original article), but for the sake of
      discussion I would like already to point out a few things that
      ‘triggered’ me.

      From what I understand Blobel claims that all the paradigms for an
      advanced EHR architecture were, already back in the nineties, defined
      in the context of the Generic Component Model (GCM) (no reference
      provided).
      In the article he states that the GCM provides a service oriented,
      model driven system architecture for the development of a sustainable
      and semantic interoperable EHR systems.
      The GCM provides a multi-model approach for EHR architectures, system
      development and implementations by the simplification of the system
      description by means of:
      - transparent domain management,
      - the composition and decomposition of the system components
      - the views from the different angles on the system (amongst which
      thorough modelling of business models

      As a result the GCM provides reference architecture for analysing,
      designing en implementation of EHR architectures, as well as a tool
      for the development of migration strategies (Educational challenges
      of health information systems’ interoperability. B. Blobel, Methods
      Inf Med 2007; 46 p.52-56)

      Although I can’t assess the article fully on it’s merits, the idea of
      a theoretical ‘meta’ reference architecture for the future which can
      be used for the purposes above seems appealing, both for further
      improvement of the openEHR architecture as well as for the future
      harmonisation of HL7 en openEHR in a common (EN/ISO 13606 derived?)
      internationally accepted EHR standard.

      So my first question is: Is the GCM to be seen as a theoretical
      ‘meta’ reference architecture, which can be used as a guideline for
      future developments? If the answer is no, why not?

      Further in the article Blobel compares GEHR against the GCM. Although
      the header of this section mentions the openEHR foundation, he
      consistently talks about GEHR and the GEHR project (MailScanner has
      detected a possible fraud attempt from "http:www.gehr.org" claiming
      to be http:www.gehr.org). The URL for GEHR links to a site, which has
      to do with different aspects of healthcare than we’re generally
      talking aboutJ). Also when Blobel talks about ADL he refers to a URL
      that doesn’t exist anymore (http://www.deepthought.com.au/) and most
      certainly it wouldn’t have linked to the latest version of the ADL

      So my second question is: is Blobel, when making his comparison,
      referring to the latest versions of the reference architecture and
      ADL as recently developed within the openEHR community?

      Blobel’s conclusion of comparing GEHR to the GCM, is that GEHR limits
      itself to the structural aspects of the knowledge components and
      doesn’t comprise behavioural aspects. Also it isn’t possible, due to
      the lack of specified rules, to aggregate archetypes. Instead they
      have to be replaced by more complex archetypes.
      More generally the GEHR approach has some essential shortcomings at
      the mathematical, system theoretical and informatics levels. These
      shortcomings have to be addressed in the future.

      In the discussion en conclusion section Blobel adds to this: that
      within the EN/ISO 13606 approach, although almost complete as far as
      semantic interoperability concerns, a lot of shortcomings and
      inconsistencies have to be solved. As example: the issue of
      structural composition and decomposition, as well as the modelling of
      business processes is not solved well.

      Personally I think that such statements should be underpinned with
      arguments/ scientific proof and/or examples or at least a reference
      to a properly peer reviewed article that does so. I would like to
      invite Blobel (and others if they feel obliged to) to provide these
      scientifically valid facts to underpin these statement, so we can
      have a proper discussion. This type of ‘review’ statements creates
      confusion, which hinders any serious discussion about future
      developments and harmonisation. It also undermines the (in my
      opinion) otherwise good intention of the article as a whole.

      My third and last question to the community is: are these conclusions
      (if applicable to the current version of openEHR) valid and if yes
      how can we address those issues?

      Cheers,

      Stef