2/18 Atlas AMPS 291, +7 44,+9 63,PMPS 47, +3 76

Status
Not open for further replies.
B

Beth & Atlas

Member Since 2010
Yesterday's Condo

I thought we might have a bounce going on as the feeder funked out on Attie. :sad: Well, this too shall pass.

I was wondering how many of you ever look at the memory in your meters. I know most meters have a function that let's you review the daily readings, but will also compile them for you as an average. I know I often refer to my meter to see the trend. When Attie is doing really well, to help determine if we need a dose adjustment based on the alternative method stated in the protocol. (Don't know what that is? Challenge one...go see if you can find it.)

And I wonder if a lot of frustrated newbies could just benefit from seeing the improvement for their cats overtime. So here is a challenge to all who read this condo today, get your meter's manual back out and figure out how to find your 7-day, 14-day, and 30-day averages...then....

Tell me your trend! ​
Attie's trend
7-202,
14-226,
30-230​
 
Re: 2/18 Atlas AMPS 291 (Trend Challenge!)

I do check it sometimes but I haven't been paying too much attention to it because I know I can't take it as an accurate average because I test so much more when he's low and that drags all the numbers down. Will be interesting to see all the trends though.

Vyktor's trend
7 - 205
14 - 202
30 - 223​

Poor Atlas, was he a bit hungry this morning, I'm sure he's over it already.
 
Re: 2/18 Atlas AMPS 291 (Trend Challenge!)

Actually, the more testing you do, the more accurate the numbers. I don't consider the amount of testing in lower numbers to skew your average as much as you think. Why? Because over the course of time those same tests are averaged into your 30 days and will still show a trend up or down.

I have seen my trend go up from a week of higher numbers, even though I am testing less. In fact, much to my consternation at times it seems the trend jumps up from just a few days.

Do you know what the n= means?

7-n=27
14-n=58
30-n=132​

So, if I am looking at my n, I average 33 tests a week. That doesn't seem to be a big variance from my recent weeks low number of tests.
 
Re: 2/18 Atlas AMPS 291 (Trend Challenge!)

I’ll use my 02/02 as an example of how skewed it gets because that provides an extreme example. The AM cycle was flat pink and I only took 3 tests whereas the PM cycle was blue and green and I took 9 tests, n is the number of tests.

If I just average the tests that I took (as the meter would) I get the following:

342 + 326 + 311 + 142 + 52 + 43 + 76 + 70 + 72 + 76 + 77 + 97 = 1,684 / 12 n = 140.33

Whereas if I had taken 9 tests in the AM (i.e. the same number as the PM) while the numbers were high I would get the following (for the extra 6 imaginary tests I am using the average of the 3 tests I did take, i.e. 342 + 326 + 311 = 979 / 3n = 326.33):
(6 x 326.33) + 1,684 = 3642 / 18n = 202.33

The 202 is much closer to his actual average BG for the day than the 140 the meter would show.
 
Re: 2/18 Atlas AMPS 291 (Trend Challenge!)

Yes, the average against any given day can be extreme, but over the course of time those extremes mellow out.
 
Re: 2/18 Atlas AMPS 291 (Trend Challenge!)

Good idea as a challenge to bring average trends to peoples attention:). It's something I've always kept an eye on, and I also enjoy looking at at overall graph of my cats progress. And happy to see Atlas is heading the right direction with the averages:).

That's a great example by Vyktor's Mum on how the frequent tests around low numbers can skew things. How much of an impact it has will depend on how frequently these days happen and how well dispersed they are. Here are 3 ways to deal with it:

1. Not worry about it. Like you (Beth) say it will average out and provided you don't have a bunch of days like Vyktors Mum's example clumped together in a week and none the rest of the month it won't impact the trend of averages. For my kitty straight up averages of all readings gives (all units mmol/l):

7- 4.18
14- 4.21
30- 4.73

This is the easiest approach and the straight up averages are still illuminating.

2. Look at averages of preshot results only. These are evenly spaced out every 12 hours and don't have the problem of a bunch of readings clumped together giving a higher weight to a low point on the curve. The spreadsheet template also makes this stat easily accessible, just average the preshot columns over the desired date range:

7- 4.98
14- 4.81
30- 5.34

(The last week was atypical for Linus and isn't directly comparable for reasons that I won't get into in your thread, he's definitely still improving though.)

3. Do some slightly heavier math. I'll skip the goriest of details, but this should ring some bells for anyone who went through first year calculus. For a continuous function like blood glucose over time, the definition of the average will involve an integral that represents the area under the curve of the graph of Glucose vs Time. Given the data points you have you can use one of the numerical methods to approximate this area. I use what amounts to a slightly modified version of the trapezoidal rule (modified so it plays better with the spreadsheet format). This effectively kills off the extra weight a low point will have from extra tests around it. For my kitty and approximation:

7- 4.32
14- 4.39
30- 4.92

Note the trend is the same as in approach 1, and this is almost surely not worth the extra effort unless you already know how to do it and have too much time on your hands;).
 
Re: 2/18 Atlas AMPS 291 (Trend Challenge!)

I do look at that a lot.

Tarragon's trend
14 day - 278
30 day - 232

Hope you are having a great day Beth!
 
Re: 2/18 Atlas AMPS 291, +7 44,+9 63,PMPS 47 (Trend Challeng

Looks like Attie is determined to work on those skewed numbers today! :lol:
 
Re: 2/18 Atlas AMPS 291, +7 44,+9 63,PMPS 47 (Trend Challeng

If KT hasn't bombed, I'll look at the 14 day but I'm like Serryn, I don't test when he's high but test a lot when he's low. Just as an example,

Feb 12 - 15x
Feb 13 - 5x
Feb 14 - 14x
Feb 15 - 6x
Feb 16 - 6x
Feb 17 - 5x

Those Feb 12 and 14 numbers will all be low bringing down those few large numbers to an artificial low average. I use the spreadsheet numbers and get 1 average number for that long range of lows then use that single number instead. Little work but gives me the info I want...

HUGS!
 
Re: 2/18 Atlas PMPS 47, +3 76 (Trend Challange)

I think the difference here is I keep saying TREND

sort of like looking at this....http://www.tillydiabetes.net/images/Lantus_Start_Graph.jpg without the graph lines.

There are artificial highs and artificial lows, but when looking at a TREND those extremes are minimized. A week of low numbers, followed by another week of low numbers, followed by another week of low numbers...sets a trend. That trend can be seen by using your meters memory.

Thanks Brian for trying to explain!!! :lol:
 
Re: 2/18 Atlas AMPS 291 (Trend Challenge!)

Melissa & Tarragon said:
I do look at that a lot.

Tarragon's trend
14 day - 278
30 day - 232

Hope you are having a great day Beth!
Click to expand...
Tarragon's SS is a good example....T went from running greens to hitting more yellows and that is reflected in T's 14 day and 30 day.
 
Status
Not open for further replies.
Back
Top