Re: Experiment on meter accuracy


"Wizzzer" <Wizzzer@xxxxxxxxxx> wrote in message
news:NyNeg.175$sB3.123@xxxxxxxxxxx
Hi Peter,

Your numbers are very interesting. Being a long timeTech, Programmer,
WAN/LAN Administrator, Systems Engineer, etc, the numbers bother me, too.
After reading your post, I was looking at my data, 104 days, since Dx Feb
14th. The more days, the more data, the screwier the numbers get. My FBG
average is 139 for the full 104 days. The highest reading is 180, the
lowest is 104. The number spread is 48 above average, 54 below and 4
right on the money. The high and average may be thrown off by 2 readings,
180 and 177, the next highest is 169, when seen in context, they are
obviously bad.

Wrong conclusion. In data that has an "expected" value you might be right
but there is no reason to believe that just because you threw a high number
it was incorrect.

The numbers have gradually decreased over the period. It would appear that
the best numbers seem to come from averaging one week at a time.

It is not about averages, it is about maximums and minimums. You wake up
and are 90 so you take 50 measurements at 90. Then you eat a ton of carbs
and blow a 450 on the meter. Obviously the 450 is wrong and the average is
90 so you are great... NOT!

That way, the weeklies, plus a running average of the dailies, plot a nice
picture of progress. I had written a program using monthly and total
averages and have decided to change that accordingly. I also have to find
a way to smooth, toss out or disregard numbers stepping too far away from
the progression.

The 8 pages of technical mumbo jumbo (labelled specifications) in my
manual are beyond belief. There is nothing in there that would give you
any idea of just how accurate or reliable that meter is. Nor is there any
warning to ignore or re-test if you encouter suspicious data. Or, even
what may be considered bad or skewed data. As one poster said, it's better
than nothing. Or is it? It's not like we're dealing with an aging outdoor
thermometer... it can be off by 10 degrees and not kill you. There is no
warning on that box, manual or unit to indicate "This unit is just a toy,
don't bet your life on it's accuracy". It would seem much more reasonable
than the decal on my mower that says "Do not place feet under mower when
blade is engaged". Just my opinion, but what the hell do I know. I'm
still a rookie, but very interested in making some sense out of these
goofy numbers.

Cheers, Wizzzer T-2 (and counting)


The accuracy is what it is. If you are 20% off you won't change the
medication by an amount to be concerned about. An extra 5 units more or
less isn't going to ruin the game, there is another test soon enough.

While you are concerned with averages remember that you don't test on
regular intervals, especially when sleeping. Also, A1C is not a linear
prediction of your actual BG average.

I think the smoothing the data and throwing away your high spikes is only
fooling yourself.
-Sax


.