How to estimate your BG meter's noise

Average BG meter error

Every BG meter has some noise in the signal it measures. In fact, what constitutes the system under measurement is [your finger + your BG meter + you], and results may vary for the same BG meter depending upon the person using it.

Here is a procedure you can use to figure out how to estimate the error you should expect to see for [your finger + your BG meter + you]

  • When you are stable around 70, take 15 consecutive BG measurements, using different fingers, trying to make a fingerpoke every 30 seconds (as fast as is reasonably possible, really, so you don’t drift too much).

  • Plot them very carefully, using a very large scale, on graph paper

  • Draw a straight line across them that averages out the signal (it may not be quite horizontal)

  • For each measurement you took, measure the difference (the “error”) between the measurement and the line

  • take the ABSOLUTE VALUE (i.e. whether it is positive or negative, use the positive value for each measurement) of each error

  • add all 15 up

  • divide the result by 15.

  • The final result is your average error around 70

  • Do the same thing when you are stable around 120. That will give you the average error around 120.

  • Do the same when you are stable around 180-200. That will give you the average error when somewhat high.

Do NOT use this procedure when you are around 40! While it would be nice to know, it is more important to get you back up ASAP! And, even if you did, your mental state is such that you would probably screw up the experiment anyway :slight_smile:

This procedure will tell you fairly precisely what error you should expect in most normal conditions.

As a note, do make sure that your fingers are clean, but don’t do your measurements any differently from how you normally fingerpoke, otherwise the results will not reflect your normal expectations.

Standard Deviation (SD)

If you want to calculate your BG meter’s SD at each level, once you have measured the errors for a BG level:

  • square each error. The result is a positive number.

  • add the squares of the 15 errors for a given stable BG level.

  • divide by 15

  • take the square root of the result

What you get is the Standard deviation of [your finger + your BG meter + you] for the stable BG level in consideration

6 Likes

Wow! I knew diabetes management was a lot to do with numbers, but had no idea you could get this detailed. Thanks for sharing!

2 Likes

The problem is that when you plot and draw a line and measure with a ruler, there is a lot of possibility for measurement error. Plus you would be eye-balling the average line.

It would be more accurate to just do it mathematically.

1 Like

I agree: I do it on Excel. But I tried to give a set of instructions anyone can do. A precise graphical job can be very accurate!

At the risk of sounding dumb, could you explain?? A line according to an eyeballed average? Or calculate and draw?

I actually think bullets 2 and 3 probably make plenty of sense once I get 1 clarified… please. :slightly_smiling_face:

It is probably easier to do this by taking the average. The line described in step 1 is just averaging all the BG readings.

This is what it is describing:

  1. Take 15 BG readings very close to one another and when flat.
  2. Get the average of those 15.
  3. Take the difference each reading had from the average. Record the positive value of it, also called the deviation. For example the difference of 6-9 is 3, not negative 3
  4. Finally, average the deviations.

If you want to post your BG measurements, I can demonstrate it for you with actual numbers. It’s very easy to do, but can be confusing trying to describe it.

2 Likes

Thank you, sir. :slight_smile: It actually doesn’t sound that hard at all. If I run into trouble, I know who to call. :smiley:

Walking that fine line between clarifying and mansplaining.

2 Likes

And doing it well. :hugs:

1 Like

Did my tests… I know I said I didn’t NEED no man’s help, but if you have some extra time and want to crunch some numbers, here. For the record, I can DO the math… I’m just checking up on your skills.

65, 64, 67, 63, 66, 64, 64, 62, 60, 60, 62, 60, 58, 60, 59

I’m testing two more meters with this… as soon as my blood replenishes.

Speaking of which, finger sticks DO hurt. Turns out the only reason they don’t hurt for me anymore is because I’ve been using the same two fingers for the last 12 years, and they no longer register pain. Or anything for that matter.

Oh, and ALSO very interesting… I DID throw one number out. But I had a good reason. I had drawn a huge drop of blood on the fifth draw. Way too much blood. It was still there when I made my way back around to that finger in the rotation. So I figured I’d just use what was there. It gave me a reading of 86… I’d heard that before, that old blood gives bad readings, but I had never seen it before… It’s all so stinking interesting. :smiley:

1 Like

Your average deviation is 2.284.

Very solid numbers. Your meter is great.

2 Likes

@Nickyghaleb what meter is this? That is a really really tight series of blood sugar measurements. Our One Touch meter doesn’t even hold a candle, i.e. we are plus/minus 25 when using it. Which is why we now use the Contour Next meter (the cool skinny one, not the grandma one as Eric calls it)

4 Likes

Some say he is no mere man…

3 Likes

I hear that when he drinks beer, he chooses Dos Equis and mutters “Stay Thirsty My Friends” under his breath just before he finishes each bottle/glass.

4 Likes

Boy you’re good. After the 63 my wife would have told me to “quit futzing around taking your blood sugar and go drink some orange juice”. I guess it’s a matter of perspective.

4 Likes

Good work. That’s just what I came up with. :grin:

Tomorrow I’ll get you meter #2…and check your skills again. :grin:

1 Like

Thank you. Thank you very much.

1 Like

This, my friend, is my most reliable and most affordable and most accessible Relion Prime. Ordered from Walmart. Test strips… $0.17 each. And this was the exact test I needed to confirm what I felt to be true… that it was WAY more reliable than my fancy and expensive One Touch.

If I have 15 strips laying around somewhere, I’m putting the One Touch to the test. I don’t think it’s gonna touch these kind of numbers.

I’m also running my Libre numbers… so far, it’s BETTER than my Guardian. It’s a frickin racket, I tell you.

1 Like

I love my ReliOn Prime. Just got in the Blu, though, and it’s spot on with it so far…we’ll see if the extra bells and whistles of the Blu make me switch.

2 Likes

We’ll compare notes. I love the price of the prime… that’s a good chunk of testing I do. Can’t affird to do that with most test strips…