David Plews wrote
> Subject: Abdominal pain algorithm
> In your message dated Sunday 14, July 1996 you wrote :
> > Trefor wrote:
>
> > >I thought that there was a computer algorithm written by someone from my part
SNIP
> > abdominal pain. I can get the reference if someone is interested
> > (McGraw-Hill, I think). I still remember that most emphasis was on history
> > and physical, not on tests. I would like to know more about that algorithm!
>
> As a HO who "had to" use the program I recall that it was based on a database of
> history and clinical examination and used pattern-matching to reach a
> probability for diagnosis. The more patient's details entered and their eventual
> outcomes, the greater the ultimate probability towards a correct diagnosis.
>
> Tim De Domabl (RIP sadly) indicated that he was working for NASA and the navy.
> The program could be used by anybody as long as they followed the questions on
> the form. As a busy HO it was easier to get the SHO to agree/disagree
> with my diagnosis!
Tim deDombal's algorithm is based on a Bayesian model of the
process of diagnosis. Its great advantage is that at each stage it
offers
* the most probable diagnoses, based on an analysis of the outcome
of the 'tests' so far applied (the test might be a straightforward
item of patient history, such as the sex or age, or it might be a
raltively complex procedure, such as a blod test). In general
several possible diagnoses might be offered, each with an estimate of
the reliability at that stage.
* a suggestion as to the most effective NEXT test, ie the test(s)
that offered the greatest differentiation among the more probable
diagnoses.
The work was originally carried out on patients suffering from
abdominal pain because this is a group for which there tends to be an
accurate attribution of the actual cause of illess, this attribution
being made after the patient had been openeded up to see what was
wrong.
I cannot remember the exact date when Tim first started his work, but
I believe it was originally done on our (Leeds) KDF9 computer in
conjunction with David Holdsworth. The KDF9 went in about 1970.
They don't build them like that any more. (Sighs and leans on stick)
Mike Wells
>
==========================================================
Professor Mike Wells
Department of Physics, The University of Leeds
Leeds LS2 9JT, United Kingdom
Phone: 0113-233-2339 E-Mail [log in to unmask]
==========================================================
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|