[Simon]

I've been plotting frequency histograms of my fasting blood glucose and the resulting histogram appears to be a double gausian.:smartass: If my fasting bg was purely random it would be a single gaussian. The presence of the extra gaussian inplies the existance of two distinct states that my insulin response can fall into. Does anyone have any idea what these represent? Sorry to be so technical but if I can isolate the two states I can reduce the variation in my bg by half!

[TAutry]

Hi Simon,

I might offer a comment if I understood any of what you posted.

What, pray tell, is a 'gausian'?

Admittedly, I am not very technical nor mathematical. I have a rather 'down home' approach to managing my diabetes: I test frequently, take my insulin, eat sensibly(most of the time) and try to get a little exercise. Even more importantly, I don't stress over it. I live, work, play and try to enjoy life.

Good luck with your 'histograms'(Whatever they are.)

[BriOnH]

I always plot my glucose levels against time and vs carbs, insulin, and exercise; I think we all do this in our heads quite frequently.

Thinking out loud here; a frequency histogram (in this case) is going to be your fasting blood glucose level population (hopefully taken at the same time everyday. how big is this population? the > the population is the more accurate you can trend) against the frequency of which you hit that particular BGL. What are your centers under those two (Gaussian) curves? Are they that far apart in respects to the variance in BGL? Assuming they are...

You have a big variance in your frequencies to achieve a trend of a double Gauss. That variance, in the end, will always rely apon one of a few critical factors. What you eat, how much insulin you take, your activity level, what your other hormones are doing. So in your split of frequencies in what you would like to see versus what you actually get will rely on those factors. Only you could guess/know (unless I saw your data, but no thanks hehe) what factors those are to normalize your data.

[jen_slc]

Do you mean you have two 'humps' on your plot? Same size/shape or do they differ in height (freq) and breadth (standard deviation)? If I understand you correctly, I would see if you can find a correlation of each hump with some other factor, like maybe one hump of bg readings is strongly associated with a certain time of day, maybe your higher hump of bg readings turns out to be mostly breakfast readings. I'm not sure anyone can really tell you what they mean other than you've got 2 humps, there could be many many factors that only you'll know about. If this is only based on a week's worth of testing (say, 5 tests/day = 35 tests), that's not a lot of data to be working with. If you try a few months, you'll get a better idea.

[DaveC426913]

Oversimplifying from a non-expert: Gaussian just means a bell curve. A single Gaussian would indicate a single factor affecting the curve. A double Gaussian would indicate two factors independently affecting the curve (it does not necessarily mean two humps).

So, I believe what he's looking for (if I may) is what factors will affect his bgs. Obviously caloric intake is the primary. Others would be exercise/activity and stress.

[rzrbks]

So, I believe what he's looking for (if I may) is what factors will affect his bgs. Obviously caloric intake is the primary. Others would be exercise/activity and stress.

You forgot the cycles of the moons of Saturn.

[BriOnH]

Originally Posted by rzrbks

You forgot the cycles of the moons of Saturn.

Intresting enough Carl Gauss (hence Gaussian) was a famous astronomer, and used most of his functions to determine where planets are relative to the earth lol.

I was actually beggining to write more about this. Is this why I dont have a girlfriend? "Your eyes are like the most beautiful double Gaussian curves I have ever seen"

[Simon]

Well if I isolate all the fasting readings after taking a fixed (10 units) dose of Lantus the night before I get means of 4.56 and 6.19 with standard deviations of 0.75 and 0.42 respectively. This is using 81 readings since July last year. Taking "snap shots" of data of a month or so at various times in the last few months produces simmilar results each time so I think I'm seeing a real double gaussian and not just a random dip in the middle of my readings.

After last night's hypo I think the relevant factor may be exercise the day before. I will need to do more research to look into this.

I did a Google on double Gaussians and blood glucose and there was some mension of the double gaussian model in blood glucose management though I couldn't find any details.

[Simon]

My investigations into early morning liver dumping (see Dawn Phenomina) have showed that what was happening was I was testing either before or after a liver dump at random each time. This produced the two gaussians I was seeing. Hopefully if I just test before doing anything else on a morning I will see a single gaussian and half the variability in my results.