Re: [Gnumed-devel] Normcurves ... still alive

[Hilmar Berger]

On 7 Aug 2002, Christof Meigen wrote:

> Hi,
> 
> Hilmar Berger <address@hidden> writes:
> > I think it would be nice to have norms not only for metric but for
> > categorial data, too (e.g. hematocrit in males / females). That might
> > facilitate age/height-independent norms in different ethnic groups or
> > patients under different conditions. Is this already possible within the
> > current design ?
> 
> What do you mean by age/height-independent norms?
Umm, I guess I didn't quite say what I meant. I just got the impression
from your code (norm.py) that there is no way to use nominal data scales 
on the x-axis and wondered if this should be covered. One could of course
split those into separate norms or translate them to ordinal scales. I
guess it's an rather unimportant implementation issue and can be handled
by what you already implemented. 

> Right now, the values that get checked by norms have to be numbers.
> But keep in mind, this covers all ordered categrories like
> status of puberty or volume of the testicles.
I referred to non-ordered, nominal categories, at least that's what I
wanted to referr to :).

> Haematocrit-Values are float numbers between 0 and 1 (or
> 0% and 100%), what's your point?
> I am not aware of a reasoable example that is not covered by this
> approach.
I guess you are right, and it's better your way.

> A different question is in how far it should be automatised that
> norms apply just to certain categories of patients, for example
> there are age/height normcurves for Turner-children etc., plus
> the whole ethnic question.

> I raised this issue a while ago with no reaction. My personal
> opinion is that the physician should choose the norm that has to 
> be applied himself, since it is impossible to cover all the
> special cases, and we don't have norms for each case, 
> like, a Prader-Labhard child with a vietnamese mother
> and a swedish father.
Agreed.

Hilmar
> 
>         Christof