Hi Wayne,
Thanks for the information! I am going to make some comments below in order to make
sure that I understand what Synapses/SynEx is about. Please let me know if my
understanding is wrong or incomplete.
---
On Mon, 11 Dec 2000 15:19:09 Wayne Wilson wrote:
...
>"This means that generic methods have to be provided for communicating
>and sharing records, as open systems in the public domain.
Are the implementations in the public domain or just the documents? I cannot find any
links to download any source code. There is a phone number listed for someone in
Italy but that seems to be a poor way to distribute public domain code.
Also, there is a "private area" that contains more documentation but password is
required:
<quote>
SynEX Architecture (final). You may see the document from the Private area (Online
doc). Document name: SynEx-Architecture(WTG& PAS-final).pdf Classification:
Presentation. 20/10/2000 *NEW
EC review on 17th October 2000 at Brussels. You may see the document from the Private
area (Online doc). Document name: GE-047 review.pdf Classification: Presentation.
20/10/2000 *NEW
</quote>
...
>SynEx:
...
>It is better for a record sharing facility to "sit on top of" the
>systems that use it, and just provide a capability of translating from
>one system into another.
This seems to describe SynEx's "Mediator Service". At first glance, it looks like
what OIO tries to do. However, upon closer examination of the "Mediator Service", it
appears to work at a system-level.
Here are its listed aims:
to connect two legacy systems
to integrate or migrate a legacy system into a SynEx server
to connect two SynEx servers
to connect a client component and a SynEx server.
In contrast, OIO's "Linkers"-metadata works at a single term and at most at the level
of a collection of terms. So with the OIO, connection between two OIO servers will
never be addressed by the "Linkers"-metadata. Instead, the target will always be
translation between two different terms on two different OIO forms.
> SynEX does have some rules, as, for instance, terminology has to be
>understood if data names and clinical terms are to mean the same thing
>to all parties.
I think you are referring to the Synapses Object Model (SynOM) and the Synapses Object
Dictionary (SynOD). For those who are interested, a description is found on page 3 of
http://www.cs.tcd.ie/synapses/public/publications/mie_aug99_v.pdf.
The SynOM is functionally similar to OIO's Forms metamodel, described here:
http://sourceforge.net/docman/display_doc.php?docid=1181&group_id=9295
There are some important differences that is not critical for this discussion but may
benefit from a more in depth discussion later.
...
>For the
>purposes of this discussion about meta-models and information exchange,
>viewpoint #3, the Information Model is where you should focus.
Regardless of the Information Model, the machinary of interest here is the "Mediator
Services". From the most detailed description that I can find, at
http://www.gesi.it/synex/component/broussais_mediator.htm - it seems pretty trivial.
The only notable feature in the design is the use of a "Intermediate Representation".
I am not so sure that this is necessary but I will reserve judgement until we get
OIO's "Linkers"-metadata working. :-)
...
>SynEx uses as it's terminology service, work done at Manchester on GALEN
>(there is supposed to be an open-galen, as well).
>
>The key thing to remember here is that Synapses and SynEx are attempting
>to federate at the system level.
Right. I (and Renner et al) think this is harder and not necessary. The system level
"federation" will be incrementally achieved through the use of the less ambitious
"Linkers"-metadata as these interchanges are needed.
>The federation architecture involves a
>common federated record architecture (European pre-standard with GEHR
>roots) which is mapped into system architectures via an object model and
>using terminology services.
Right. In my view, this requirement for a "common federated record architecture" is
the Archilles heel of the entire infrastructure. This is almost the same as saying:
if we agree to a standard, then we can inter-operate.
OIO is different in that it requires no pre-agreement on metamodel or schema and
allows the building of "mediators" as needed. Although if two systems use the same
metadata (e.g. two OIO servers that host the same OIO form), then a trivial mediator
that maps data by item name is sufficient (this is already implemented in OIO-0.9.4).
For two system that use the same metamodel but different metadata (i.e. two different
forms), having a mediator (="Linkers"-metadata) will be adequate. For two systems
that do not use the same metamodel (e.g. FreeMed and OIO), then it is necessary to
first create an OIO Form that maps to the foreign metadata, import foreign data into
the OIO Form, and then use "Linkers"-metadata.
> OIO will need to federate at the level of the individual clinician.
Yes and No. OIO is _capable_ of federating at any level - but it is highly unlikely
that there will be no sharing of "OIO Forms" between clinicians. In fact, the whole
idea of OIO Forms (and other metadata) Libraries is to encourage sharing and reuse of
metadata components.
For example, an insurer or specialty group may suggest or mandate the use of certain
forms. OIO allows individual clinicians to use additional or different forms - but
does not mandate what clinicians use one way or another.
>This is quite a jump in scale and I am not sure the SynEx architecture
>is up to it. To quote SynEx "This is not case-by-case integration.
>With case-by-case solutions, linkages tend to increase geometrically as
>more different systems are connected, and soon become unmanageable.
Again - the designers of SynEx did not have the benefit of reading Renner et al, 1996.
Even when a solution is capable of "case-by-case" linkage/mediation, it does not mean
that all cases must be connected to all other cases! Again, this is a radical step to
take but I think a very important one.
> A
>generic method of federating records (in the engineering sense of
>federating), such as SynEX, becomes progressively simpler and remains
>maintainable and manageable."
Right. In the special case where the set of "Forms" is small, OIO performs like
SynEx. I am not sure from the description whether SynEx's Mediator Services allow
plug-and-play "mediators" and whether they are at the same granularity as the OIO's
"Linkers"-metadata? What we will have for the OIO is an online library for these
"Linkers"-metadata similar to the current OIO Forms Library.
thanks in advance,
Andrew
---
Andrew P. Ho, M.D.
OIO: Open Infrastructure for Outcomes
www.TxOutcome.Org
Assistant Clinical Professor
Department of Psychiatry, Harbor-UCLA Medical Center
University of California, Los Angeles
Join 18 million Eudora users by signing up for a free Eudora Web-Mail account at
http://www.eudoramail.com
