> Agree with Eliot. We could also apply "learning" so that what's initially
> loaded are:
>
> 1. the MD's commonly prescribed drugs for that ICD diagnosis
I know that the gods have put the diagnosis in front of the therapy, but in my country
(Oz) doctors are used to chose a drug, and after that to be asked by their "favourite"
software: "what are you prescribing that for?". Chance of a snowball in hell to change
this habit. It has historical reasons, as here medical computing evolved out of
computerized prescription packages - computerized progress notes came much, much later.
> 2. followed by other MD's commonly prescribed drugs for that diagnosis
> 3. followed by MD's commonly prescribed outside of that diagnosis
> 4. commonly prescribed drugs by all MD's by frequency
Anyway, even if we could change this nasty habit, this workaround would only work for
the drugs, not for the patient's pick list. Haven't got a list of "commonly coming
patients", they come randomly (apart from the 60% pre-booked ones, but then I've got
probably more disruptive phone calls with completely unpredictable patients calling
each day. And here I need INSTANT response time by my software, as I do NOT get paid
for answering these phone calls in my country. Curse on Medicare.).
> Can anyone please comment on this sequence (I just wrote this off the top
> of my head).
I think the item most likely to reduce the problem would be the list of the "commonly
prescribed drugs" by the logged in doctor, as most doctors have only a limited
repertoire of drugs they usually use fro the vast number they could use. But again,
willnot work for my patient's names list.
Thanks anyway,
Horst
