At Mon, 25 Nov 2002 13:32:12 -0500 [EMAIL PROTECTED] wrote: >I wonder whether access to large federated data stores of recent >symtoms and diagnosis would have caught her GB on the first ER visit?
You are not the first to wonder... Larry Weed (of Problem Oriented Medical Record fame) was working on a clinical tool to do this which he calls a problem-knowledge coupler. I did look at the software once: you enter a symptom, it conjures all kinds of unlikely diagnoses. The physician who I interviewed who was using it tended to use it as a fallback, in the way one might use a book like French's index of differential diagnosis. And the HELPP system at LDS in Salt Lake is an example of a rule-based system that supports antibiotic prescribing based on the prevalence of sensitivities recorded in the hospital lab. I am sure there are many others. But the specific clinical situation you describe is a very complex problem indeed, mainly because clinical data is not gathered in a linear fashion by anyone much beyond junior clinical student level. I think we may be relying on clinical intuition (and followup visits) to catch those trickier diagnoses for a while yet... with all that it implies, both good, and bad. Regards to all, D. -- Douglas Carnall +44 (0) 207 241 1255 +44 (0) 7900 212 881 http://carnall.org
