Re: [Gnumed-devel] Measurement workflows - value colour red , display of � , and gnuplot auto-invocation?

[Karsten Hilbert]

On Sun, Jul 14, 2013 at 01:37:11AM +0000, Jim Busser wrote:

> >> In GNUmed, clinical importance ("relevant") and the
> >> numeric or text value are both combined into red characters.
> > 
> > Apart from preference I fail to see why that is a problem ?
> 
> Perhaps it is just me, and perhaps it is just by force of habit (but I doubt 
> it) ...
> 
> … for any one test, when I visually scan across the line of results, and I 
> see changing numeric values, it is easier for me to focus on the numeric 
> magnitudes, in other words "how much did these numbers change or fluctuate, 
> and in what direction) if the characters are all in the same font and colour.
> 
> If normals and abnormals are in different fonts and/or colours, it can become 
> too distracting.
> 
> In the case of a series of values of a single test, all of which would be 
> (say) low, it is easier for me to observe, on the basis of the cell shading, 
> that the entire set is low, after which I can mentally "suppress" the 
> background shading to better focus on the numbers themselves. For example, to 
> decide whether there are trends within the block and then, if the field of 
> view includes the values to become normal again, to see whether only 
> marginally or fully normal.
> 
> Take the example of a patient who is delirious, and who was not (and still 
> does not) eat or drink very well, whose sodium (Natrium, Na) while on 
> intravenous fluids had been low, and later after discontinuing intravenous 
> fluid becomes highish.
> 
> If a series with the most recent value is shown at left, thusly:
> 
>       145 <--- 142 <--- 139 <--- 136 <--- 133 <--- 130 <--- 127 <--- 124 <--- 
> 122 <--- 121
> 
> then I personally  find something advantageous in a shaded background, 
> because by shading the background of cells (containing values) 133 through 
> 121 the abnormal state of sodium is more-effectively communicated to me as 
> (in all likelihood) continuously low for that entire block of time, since the 
> shading would be contiguous / continuous across all those values.
> 
> If, instead, the shading were to be uniform, as is currently the case in 
> GNUmed, then it would only be the characters themselves "130" etc which would 
> appear "abnormal".
> 
> To me, this does not communicate as effectively.
> 
> It also, for me, becomes easier to "focus" on "just the numeric values" when 
> all of them are of the same font colour. I can then avoid to feel distracted 
> by the visual transition from "red" 133 to "black" 136 because even though 
> the 136 is "normal" and the 145 are "normal", the clinical process does not 
> care about such dividing lines.
> 
> I am better able to see, when I can better ignore (mentally suppress) the 
> shading, that not only has the patient gone from subnormal to normal, but we 
> seem to be witnessing a new problem, where the patient is not drinking enough 
> free water.
> 
> I realize that I am providing an inpatient example and that the focus of 
> GNUmed is outpatients however I think the principles of how information 
> communicates to us remains the same. I also realize that some clinical 
> systems highlight clinically significant changes even within normal ranges, 
> and might therefore flag the patient whose sodium changes from 136 to 145 
> within 3-5 days but that is a separate issue.

OK, I see what you are getting at. It sounds quite
reasonable to me. Please suggest colors (and ways to
communicate what they mean).

Karsten
-- 
GPG key ID E4071346 @ gpg-keyserver.de
E167 67FD A291 2BEA 73BD  4537 78B9 A9F9 E407 1346