I think this community looks very interesting byt need a crash cource in order to learn the "tribe language" and what open EHR is all about.
I am a physician and has hence approached the field of medical informatics and medical records standardisation from the clinical wiew.
Arild
I have to agree with you - but please realise that the time is FAST
approaching when much of the debate - certainly the majority - will be in
the clinical area.
The important advance that seems promising in the design of the openEHR
approach is the ability to model clinical concepts (as archetypes) that we
want unambiguously represented in the record. This appears quite
straightforward initially - until you consider that we would like computer
systems to understand these concepts even when they are entered in various
different systems.
The openEHR reference model looks after data interoperability and assures
clinicians and patients that their record can be read and stored (without
losing information) at any site This is already taken care of.
The next issue is what I call 'Clinical interoperability' - that is we need
to share information based on ideas and concepts that we share - this is not
too much to ask and we have achieved it to some extent on paper. I can go to
most hospitals and find my way around the records - the parts are usually
labelled and layed out in a manner with which I am somewhat familiar. What
is usually most difficult is to find out what are the fundamental problems
facing the patient. In the EHR this is about navigation and useful data
collections based on best practice. I would suggest that this is largely
organisational and is best served by being able to locate and present
information to the clinician reliably. 'Organisational' models are required
to be explicitly understood in a knowledge base and for this purpose we have
the organiser archetypes in openEHR. We need to know more about how to
generate these in a way that they can be related to one another - so
different clinicians can look at the record as it was created - as well as
in forms that suit them and their work.
The holy grail is 'semantic interoperability' so that the computer can read
and understand the information as well as us. This requires that the
information is recorded in a way that ensures that important clinical
concepts are represented in the record unambiguously - and if we are to stay
sane - that there are a limit to the number of useful concepts. Don't forget
that we have terminology to populate these 'useful concepts' further. So we
can have one archetype for Problem/Diagnosis/Histological diagnosis - using
specialisation to deal with added features and dropping optional ones that
do not apply. This is the world of the primary or entry archetypes - clear
expressions of useful clinical concepts - both clinically and for automatic
processing.
Any questions and where do we go next?
Cheers, Sam
I would suggest that any medical recording system that insists on
limiting the number of useful concepts will be fatally flawed to the
degree to which it limits that expression. For example, the expressions
"cat died" and "bereavement reaction" in a patient's notes could be both
the same thing and a world apart.
Of course, for the big, barn-door pathological diagnoses we can (and
should) aim for unambiguous expression. But even there there are major
problems: for example, temporal relations alter our perspective on
events, and cause us to express ourselves in different ways, and with
different requirements for precision.
The "quoted" words represent the same incident:
"the onset of crushing central chest pain caused by a left anterior
descending acute coronary syndrome" in which the specialist weighs up the
potential risks and benefits of thrombolysis vs angioplasty" (60mins from
onset)
"MI 1998" which is sufficient evidence for the GP in 2003 to continue
aspirin. (5 years from onset)
I can't help thinking that the contributions to any medical record need
to be as structured/formless as the users will agree for them to be at
the time, and I am very wary of any system which claims to impose this by
software fiat. Even in the narrow world of the technical discourse of
doctors the "importance" of a concept is a subjective matter.
D.
Doug
>The holy grail is 'semantic interoperability' so that the
computer can read
>and understand the information as well as us. This requires that the
>information is recorded in a way that ensures that important clinical
>concepts are represented in the record unambiguously - and if we
are to stay
>sane - that there are a limit to the number of useful concepts.I would suggest that any medical recording system that insists on
limiting the number of useful concepts will be fatally flawed to the
degree to which it limits that expression. For example, the expressions
"cat died" and "bereavement reaction" in a patient's notes could be both
the same thing and a world apart.
I am not saying that we want to limit the number of concepts - just that we
need concepts that are useful to share. I am not talking about the content
itself but the clinical concepts that contain this content - the archetypes.
Clearly "cat died" is a report of an event - perhaps the patient's story -
bereavent reaction - is some king of assessment of the situation. Don't
forget that the concepts that I am talking about are the constructs that we
want to share unambiguously.
Of course, for the big, barn-door pathological diagnoses we can (and
should) aim for unambiguous expression. But even there there are major
problems: for example, temporal relations alter our perspective on
events, and cause us to express ourselves in different ways, and with
different requirements for precision.
Certainly.
The "quoted" words represent the same incident:
"the onset of crushing central chest pain caused by a left anterior
descending acute coronary syndrome" in which the specialist weighs up the
potential risks and benefits of thrombolysis vs angioplasty" (60mins from
onset)
A useful concept here is the recording of the acute event - this is
different than the recording of information which is considered persistent
over a long period.
"MI 1998" which is sufficient evidence for the GP in 2003 to continue
aspirin. (5 years from onset)
I agree
I can't help thinking that the contributions to any medical record need
to be as structured/formless as the users will agree for them to be at
the time, and I am very wary of any system which claims to impose this by
software fiat. Even in the narrow world of the technical discourse of
doctors the "importance" of a concept is a subjective matter.
Again, I think you are talking about content and what archetypes offer is
the ability to mix these things as the clinician feels is appropriate.
Please stay wary..
Cheers, Sam Heard