On 26 September 2011 13:33, Saul Galloway <[log in to unmask]> wrote:
> A long time ago in Sheffield Childrens Hospital there was a DOS based
> program for assessing acute abdomens. You input temp, WCC, where(if)
> they were tender etc and it output a differential
>
> 60% nonspecific abdo pain
> 28% mesenteric adenitis
> 11% appendicitis
> 1% velluvial matrix deficiency syndrome
>
> Even if the predictive algorithm was entirely accurate, what % chance
> of acute appendicitis are you happy to not operate upon?
Based I think on the work of Tim de Dombal at Leeds, where his fellow
surgeons were doubtful for the reasons you suggest.
But the point was made that giving more clinical information to such
programs tends to improve the quality of the diagnosis, whereas giving
more information to human diagnosticians tends to cause confusion.
--
Michael Leuty
Nottingham, UK
|