Beeminding Blood Glucose

A Beeminder user who now works at an insurance company sent me a continuous blood glucose monitor (CGM). Exciting! I’ve been wearing it for a few days and of course it’s fun to watch my blood sugar rise and fall when I eat, but what I really want to figure out is whether this metric is beemindable!


(My goals include staying at my current weight, health and longevity and all the obvious things, and eating as much candy as possible without inducing diabetes. I’m not entirely kidding, I have legit positive utility for candy. Don’t judge me!)

Here’s how the app looks right now, btw, after fasting all day:

It’s a Dexcom G6 and involves wearing a patch on your skin that talks to an app on your phone. The patch is 100% not noticeable [PS: I meant you don’t feel it at all and is waterproof etc – it’s definitely visibly noticeable]. I guess technically it’s stabbing you with a very tiny needle but applying it didn’t hurt at all. It’s really impressive tech!


congratulations, you have become a few % more cyborg :stuck_out_tongue:


That’s marvellous, and I so want one :slight_smile: Peter Attia MDhas a bunch of stuff on these on his blog and podcasts, albeit mostly for IF and ketosis measures. Consencus says very useful to measure peak levels after eating, which appears to be one of the critical numbers.

1 Like

I was in a study with a glucose monitor. I had a pet hypothesis that the world would be better if everyone knew their glucose at all times, to help avoid low glucose grumpiness. While my monitor was not the most advanced, I was really surprised by how well my insulin response worked, even after a few decades eating a lot of sugary garbage. I didn’t really get anything actionable out of it, unfortunately.

I do want to try the experiment to see if I can get anything actionable out of a more advanced monitor. Have fun and good luck!

1 Like

I’m diabetic. That’s the same CGM I have.

My first use of Beeminder (which was my only use for a few years) was beeminding how many times I tested my blood sugar (I didn’t have a CGM then).

One of my earliest interactions with customer support involved you specifically telling me this exact thing was a bad idea (reformatted for clarity):

our general feeling is that beeminder works best for things that are
under your direct control, like minutes of exercise and the like.
weight is pushing it but we’ve done a lot of work to make that viable
to beemind, with the auto-widening road and taking the min weight of
the day and whatnot.
i don’t think bodyfat works well, for example.

Okay – blood tests is a lot like minutes of exercise. The only significant
difference is that I can’t do a bunch of them in one sitting, which I don’t
think is terribly relevant. So it sounds like Beeminder ought to work for

and, yes, i had actually thought you meant beeminding your actual
blood glucose levels. beeminding how many times you do the blood test
is absolutely perfect for beeminding.

If you’ve changed your mind then carry on, just seemed worth pointing out (personally I still agree with dreev-2013’s reasoning though).


Great point, @kenoubi and past-dreev! (And as a diabetic you’d want to be especially careful that beeminding blood glucose didn’t ever induce you to do something medically contraindicated.)

I’ve gradually gotten more pro-beeminding-scale-weight since 2013, though there are a lot of important caveats that I’m being exhorted to consider when recommending that, which I mostly don’t. So I generally still agree: better to beemind things you have direct, unambiguous control over.

Lowering the bar a bit, setting beemindability aside, what’s even actionable for non-diabetics in knowing their real-time blood glucose?

Ie, how can I use this data to be healthier? Minimize the variance by eating smaller portions more frequently and/or by avoiding things with high glycemic index? And how much does that really matter for non-diabetics? And how does nutrition even work?

If yes to keeping my blood sugar as steady as possible, is literal variance (mean of the squared deviations from the mean) the best metric or do the peaks/troughs matter more?

You mostly don’t consider the caveats? Or you mostly don’t recommend it?

Me too!

As an extreme example (a reductio ad absurdum of this kind of thing) imagine installing a CO2 detector in your lungs and Beeminding CO2 levels to tell you when to breathe.

For non-diabetics, your body has a perfectly good control system for monitoring blood glucose, which is to signal hunger when your blood sugar gets too low, and to release insulin when it gets too high.

The only thing there is to do is to make sure that this system works properly, but I think that’s better done through regular doctor’s appointments to test for diabetes if indicated, rather than continuously monitoring your blood sugar level.

A non-diabetic could try to do a “better” job of regulating glucose by relying on the measurements, rather than natural signals, for when to eat, and injecting exogenous insulin - my guess is that doing so is unhealthy and could lead to disabling your body’s natural ability to regulate, in much the same way that, for instance, taking exogenous steroids can harm your body’s natural production of hormones.

I have similar concerns about beeminding weight, as well - it’s also not very much in your control, and your body also has a natural control system to regulate it (hunger and satiety).

I don’t know how it’s actionable, but the reasonable goals given CGM data are 1) never have “low” blood sugar, defined as <60 or when you start to feel symptoms, whichever comes first (contrary to popular opinion, this isn’t a big problem; hypoglycemia is actually quite rare, and by far the most common cause of actual hypoglycemia is diabetics overtreating themselves), and 2) minimize area under curve given 1. More sugar exposure is more bad in a reasonably linear fashion (until you get to diabetic levels, at which point it starts to dehydrate you and possibly other nasty things).

1 Like

I think it was clear in my original: I mostly don’t recommend beeminding weight. I’ve just become less and less shy about describing what works for me personally. Maybe that sometimes crosses the line to endorsing it as a strategy others can use. I’m thinking extremely hard about the caveats, as you know.

I’m pretty sure no non-diabetic is considering exogenous insulin.

Content warning: weight loss

Weight loss debate is off-topic here (my fault for broaching it!) so let me give the briefest possible reply to @zedmango and then anyone who wants to debate it further, DM me for access to an ongoing semi-private weight loss debate thread. (Seriously, I’m invoking moderator superpowers: No further weight loss debate here!) Brief reply: @zedmango, you’re super wrong! Ok, presumably right about yourself – your hunger and satiety signals suffice to regulate your weight – and perhaps right more generally in that with proper diet and exercise, hunger and satiety signals can suffice to regulate weight for many/most people. But for me, my hunger and satiety signals do not adequately regulate my weight and my weight will increase without bound if I rely on them. That’s an empirical fact about me and my current environment. I may be better off manipulating my environment to fix my hunger/satiety signals but I personally feel ok about simply overriding those signals, eg, by doing intermittent fasting.

Ah, perfect! And another way to say “area under the curve” is simply the average daily reading, right?

They’re not the same mathematically, but yes, optimizing for one is the same as optimizing for the other.

1 Like

Right, I guess the area under the curve is the day’s average times 24 hours, and that constant factor doesn’t matter. Mean value theorem FTW!

So now I’m wondering if that metric could be perverse. Could I minimize it by eating nothing but one big bowl of jelly beans per day for a single spike that’s as narrow as possible?

1 Like

How long have you had the glucose monitor on, and you haven’t tried eating a big bowl of jelly beans yet?


I actually think that would maximize it, because your body takes time to break down all that glucose, and you’d get a triangle of height 3h as it gradually goes down, whereas if you broke it into three smaller bowls and ate one every 8 hours, you’d have three triangles of height h each, which is one-third the total area.

This is day 4 and I’ve had only small handfuls of jellybeans (since I beemind them).

1 Like

We need you to eat a big bowl and graph what it does to your glucose! It’s for science!


I’ve literally been trying to get a hold of a CGM for over a year. Every doctor tells me to take a hike even though I am entirely willing to pay out of pocket. Any advice on how to procure one from someone who has had more success?

There are so many interesting questions I want answered. What does xyz food do to my body? How quickly does working out drop my blood sugar? Are certain types of workouts more effective? Plus many more.


Not really. Your body regulates blood glucose pretty tightly.

I think you could just take a week of data collection without any goals and then use that to set a floor, though. Technically keeping your blood glucose at 62 (assuming you feel fine) might be best from the perspective of minimizing damage to your body from glucose, but it would be awfully hard to eat a reasonable diet and achieve that.

I’m also mystified by @dreev being given one of these “just because”. Since I’m a type 1 diabetic with good insurance, I don’t have to pay anything, but my understanding is that these things are absurdly expensive. Thousands of dollars a year for the sensors.


If you wanted a metric I would choose something like minutes in the day where you are within a certain range… buuut even with my craziness of wanting to track everything about my body id be really worried about commiting to any metric. not sure that metric would really be directly actionable though it may be a good proxy for glucose related damage you’re incurring.

My personal plan was to just use it to answer questions about how my body processes certain foods and responds to exercise.

The dexcom he has is expensive. It’s 300 month at least if you pay out of pocket. I’ve been trying to get a prescription for the freestyle libre which is a 14 day sensor you can read with your phone. It’s only 70 per sensor, so about half the price. I’d probably want to record a couple months worth of data and that’s it.

Rant incoming: it is honestly pretty frustrating that I can’t convince a doctor to prescribe something that is not a drug and minimally invasive. The biggest risk is I get an infection at the injection site. Moreover there are tons of docs that just wear them for info purposes as well. The gatekeeping leaves me feeling so condescended to every time I ask. Like a child begging a parent for ice cream. The paternalistic finger wagging I’ve encountered is such a strange contradiction to all the other areas of life where we are admonished to take more personal responsibility.

I will probably be forced to have one shipped to an EU address because they are sold there without prescription. And then find some way to bring it back.


It is really upsetting when doctors are so paternalistic. You have to drop doctors who don’t treat you with respect and find ones who do. A doc who leaves you feeling like that is terrible and should be avoided. Having high standards for docs is really important. Keep trying more docs and don’t give up.

What are they saying to you? I don’t understand how they could even object.

You won’t have any trouble bringing it back. I’ve been to Europe a dozen or so times with mine and never gotten any questions in either direction.