>Why are we not using them to sort patients into those who deserve a
>second look because they are sort of like someone who had a bad outcome
>recently?
We are, aka the experienced GP. Even on a good day I'm just a well paid
pattern recognition engine, and most of my data inputs are unconscious. I
used to worry about my rule set getting fuzzier and fuzzier but more and
more I think (hope) it's an inevitable function of volume of data (in this
context patient consultations). At it's most extreme this gets as
inexplicable as the consultant that asks for a serum calcium on patient
presenting with headache, because he once saw another bloke called Pettifer
about the same age with a similar thing and he had a stutter too.
Sometimes there are spectacular mismatches. If you feed in :
TATT
Insomnia
Irritable
Tense
Marital difficulties
Errectile dysfunction
Work stress
into my algorithm (actually this makes me wonder if verbal consultations
couldn't be deconstructed into key words just like a spam filter)I'm up a
ladder of inference so high I can't see the ground, which of course kicks
into play a particular "filter" of expectation.
About 30 years ago at Sheffield Childrens Hospital they were trying to
design an algorithm to help diagnosing the acute abdomen. You fed in the
WCC, temp, ticked the box for aspects of history etc. and the system as
supposed to learn from the outcome and recursively (I think) weight the
presenting features.
What you got though was a differential in the form
69% mesenteric adenitis
20% functional
11% appendicitis
which was surprisingly little use in the decision of whether or not to do an
appendectomy. The bit that would be could would be the "this symptom
combination is 90% likely to be <commonish condition> as you correctly
diagnosed, but please also consider the 0.01% chance of <Von Eponymausens
syndrome> which commonly has the feature <indentifier>
***********************************************************************
This message may contain confidential and privileged information.
If you are not the intended recipient you should not disclose, copy
or distribute information in this e-mail or take any action in reliance
on its contents. To do so is strictly prohibited and may be unlawful.
Please inform the sender that this message has gone astray before
deleting it. Thank you.
2008 marks the 60th anniversary of the NHS. It's an opportunity to pay
tribute to the NHS staff and volunteers who help shape the service, and
celebrate their achievements.
If you work for the NHS and would like an NHSmail email account, go
to: www.connectingforhealth.nhs.uk/nhsmail
***********************************************************************
|