You go away for two days and come back to find a raging debate.

Reason for being away was attending a seminar on SGML in health care
and listening to Tom Lincoln - who has been a leader in the US in this
area - amongst other speakers.

Very interesting stuff - (and nice to meet you Gerard - hope you had a
safe trip home). When I get a chance I am hoping to try and pull all
the pointers to the various resources together, but it'll take a bit
of time.



Anyway amongst the 79 emails, Jon Rogers said

> I can't see how publishing lab results on the NHSNet in
> HTML format, without a structure, and it is reliably interpreting
> that structure which is the hard part of EDI.

As Andrzej Glowinski pointed out later, the idea that HTML isn't
structured is incorrect. It's as structured as defined by the
standards, which in HTML's case *may* be a little limiting. But if one
moves out to SGML, then there's even more freedom to define both
structured and unstructured components of a document aka medical
record and there is work underway on HCML - Health Care Markup
Language - as an instantiation of SGML (just like HTML is a subset of
SGML) but targetted at the specific characteristics of healthcare.

What is attractive is that one can start with the bits you can
structure now and leave the other bits 'til later, evolving as you go

Also the whole issue of context is one which is well recognised by
SGML proponents. It's hard, but IMHO no harder than any other way of
defining structures.



Jon continued

>How would you generate a graph of Hb results from HTML?  
 
It's not just text which is being investigated. In the US there are
apparently demonstrable voice recognition systems and encounter
tablets using stylus input which can be used to capture data and drop
voice/clinical diagram fragments into SGML structured documents. They
are simply other objects - and they are going in to SGML databases.

Of course there are problems - performance of the backend database may
be one, but that may improve, or one could implement the SGML
structures using different sorts of database technologies for speed
(M?), but retaining the SGML model.




Now over to John Williams who said

>> Meanwhile, let's also sort out what Web based records
>> really can offer and plan to implement something when we are sure
>> what we want, and how to do it

Ahmed replies

>Now that's very sensible John.  Would IMG please fund a pilot study?

Part of the seminar was to hear what had happened in a small
IMG-supported exploratory investigation into the application of SGML
in medical records......



Time for bed....



___________________________________________________________________________
John Farenden - Secta 
Triton House, Hare Park Lane, Liversedge, West Yorkshire, England, WF15 8HN
Tel 01274 852160, Fax 01274 852159
Member of the British Medical Internet Association


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%