Anyone tried January.AI?

[Ron Put]

I am tempted, as CGMs are not available without prescription in the US. I was going to get a prescription, but for $288 January.AI is rather comparable to the monthly cost of the CGM alone, and it appears to provide a pretty good app from what I can tell.

Did anyone here try it?

[Ron Put]

Well, I am trying it. It's a way to get the first-gen Libre CGM without a prescription in the US. The price is reasonable ($288 with the current discount) and it includes two sensors and an app that hopefully will provide some insights. I've spent three days with it so far.

On all three days, I ate mostly at restaurants and mostly food I don't normally eat. It included Indian (including white rice, and bread), Italian (including pasta with pesto, bread), and American (including breaded, fried Brussels sprouts, a sugary cocktail, and a whole mushroom pizza without cheese). I figured this is kind of a trial by fire (or refined carbs). I was pleased that my glucose never went over 130, despite all the refined carbs.

What I am less pleased about is that my average glucose is 103. This translates to A1C of about 5.2, so it's not widely off the mark. My A1C is usually about 5.

My lab blood test of about 10 days ago shows fasting glucose of 82 (my insulin result was 3.9, so I am unlikely to be insulin resistant). So far, the Libre patch has not gone this low, generally ranging between 95 and 115, with occasionally reaching the low 80s and the high 120s.

Here is a screenshot from LibreLink (fasted until about 15:30 yesterday, ended up with the munchies close to midnight and ate 150g or so of Black Garlic with almonds and walnuts...😞)



I've set the desirable band to be 70-130 and I was 77% within it. I noticed that Kambucha seems to cause a spike to about 120 or so.

I am still waiting for the January.ai app to gather enough data, but here is an example of an insight it offered:



Note that the "garlic" is actually Black Garlic, which is rather sweet and I expected it to spike up, but it actually caused a dip.

I am pretty happy so far with the app and the info, although who knows what the sensor accuracy is.

[mccoy]

If your lab fasting glucose was 82 (a good value by the way) I would expect a more strict adherence to it by the sensor, unless the lab test was referred to an anomalous day. More clearly speaking, the sensor reading 95 as a fasting value seems too far off the 'real' value. In freelibre I noticed that at the beginning the sensor was closer to the classic strip test value (which itself has an error which may not be negligible) but after 10 days it drifted upwards and I had to construe the results accordingly.

 

The interpretation of peaks/troughs is also important, although it should be correlated to the ingested meals, I wonder why the graph you posted exhibits so many peaks and slumps, from what you describe you don't eat so many meals...

 

Edited June 22, 2022 by mccoy

[Ron Put]

13 hours ago, mccoy said:

If your lab fasting glucose was 82 (a good value by the way) I would expect a more strict adherence to it by the sensor, unless the lab test was referred to an anomalous day. More clearly speaking, the sensor reading 95 as a fasting value seems too far off the 'real' value. In freelibre I noticed that at the beginning the sensor was closer to the classic strip test value (which itself has an error which may not be negligible) but after 10 days it drifted upwards and I had to construe the results accordingly.

The interpretation of peaks/troughs is also important, although it should be correlated to the ingested meals, I wonder why the graph you posted exhibits so many peaks and slumps, from what you describe you don't eat so many meals...

 

I am not sure about the accuracy, since I don't have a way to test blood. The sensor may be reading a bit higher, based on A1C, and my fasting blood glucose is usually within the 80s in lab tests.

But as you point out, it's the curve and pattern that's more interesting. On the first screenshot in my previous post, the peak between 15:00 and 18:00 was a somewhat strenuous hour-long hike/trail run, which pushed me to 131 at its peak. The following peaks are pizza, cocktails and stuff like that, followed by a late-night munchies snack. Mostly junk food/refined carbs, and I was sort of happy that it never pushed me over 130. I am not insulin resistant (my last insulting lab test was 3.9, at the same time as the fasting glucose of 82).

But yesterday I did a glucose challenge, which pushed me way over. I fasted until about 15:00 (so 15 hours or close), with black coffee and green tea as the only consumables. At about 15:00 I drank a solution of 75g of glucose provided by January.ai. Orange, nasty sweet-tasting liquid 🙂

In an hour I peaked to 174, then by the second hour I was down in the 110s and I crashed, hitting 50s about three and a half hours later, still fasting until then. At that point I ate and then dosed off until about 22:00 (I never just dose off). When I woke up I ate some more and continued snacking until about 01:00.

In other words, the glucose challenge really screwed me up 😞  Here is a screenshot of what happened during the first 3 and a half hours after the glucose drink, from the LibreLink app:

 

And here is a longer period that shows the crash after 3 and a half hours, and then bounces from food intake and probably my body trying to recover from the glucose challenge:

Interestingly, my Garmin watch registered noticeably increased stress (measured mostly from HRV, I think) as soon as I drank the glucose. It's similar to what I see if I drink alcohol. Here is a shot, the orange is the stress after the glucose:

I don't remember if you did a glucose challenge while using the sensor, mccoy. If so, what was your experience?
 

Edited June 22, 2022 by Ron Put

[mccoy]

No, I didn't, after installing the sensor I just hated seeing peaks!

I just remember the ADA guidelines, that during a glucose challenge after 2 hours of ingesting the solution blood glucose should be below 140 mg/dL. Strangely, in the guidelines they ignore all the rest. 

One time I did something similar to the glucose challenge, when I was pretty thirsty and drank cereal milk, unsweetened, which was cold and refreshing. I drank the whole one liter (2,2 pints) of it, and after a while I felt a little dizzy and sweaty. A strip test revealed a slump of about 70 mg/dL, a low value I never registered before on myself (it might have been lower).

That's inevitable when ingesting an abundance of liquid sugars, it was enough to discourage me for the future to drink anything sweet.

[Ron Put]

On 6/22/2022 at 1:24 PM, mccoy said:

No, I didn't, after installing the sensor I just hated seeing peaks!

I just remember the ADA guidelines, that during a glucose challenge after 2 hours of ingesting the solution blood glucose should be below 140 mg/dL. Strangely, in the guidelines they ignore all the rest. 

I have the feeling that the high-fat/low-carb crowd focuses on flat lines too much, at least in part to sell services and products. It's what virtually all outfits that sell sensor subscriptions do, because it's easy to point to peaks and justify the need for the service and the products.

Of course, glucose is only one part and I am not sure how valuable it is in the end, without taking insulin into account. I have a pretty high carb intake, averaging 61% of my diet over the past year, with 14% protein and 24% fat, according to Cronometer. Most of the carbs come from chickpeas and other legumes, as well as sweet potatoes and to a lesser extent grains such as steel-cut oats, dark sorghum, and flaxseed. Most of the protein also comes from legumes, and also from tofu, mushrooms (and tomatoes?!). I wish I could go down to 10% of protein, but it seems too hard for me.

I noticed glucose peaks as high as 150 when I eat a particularly large meal. Usually, it's legumes, sweet potatoes/yams, crushed tomatoes and greens, but the size appears to be of particular importance. Such peaks usually occur around the first hour and then I am back to around 95-100 about 2 hours later, sometimes 2.5 hours.

I wonder if others here who eat very large meals at once, like Michael or Dean, also spike depending on meal size?

But the bottom line is that such spies appear to be perfectly normal, as long as there is good insulin sensitivity. Those on high-fat and high-protein diets may spike less as long as they stay away from carbs, but then their insulin sensitivity becomes impaired, resulting in large spikes after relatively minimal carb intake, which to me is not a good thing.

January.ai may be more useful to people who are less aware of their diet, but to me, it's pretty useless. I feel the same about all the other similar outfits, such as Zoe, NutriSense, and Levels. Basically, the whole "personalized nutrition" business model. They make flatlining the glucose curve a sport, but I don't see any real scientific evidence that for healthy people it is beneficial, and in fact, they can encourage insulin resistance by promoting high-protein, high-fat diets.

The AI part is grossly overstated, basically showing comparisons between meals and the resulting glucose curves. Pretty primitive and simple. They require a Fitbit or Apple Watch but appear to only take HR  data from it, which is not all that useful. They don't even utilize sleep and exercise data from the tracker, and both affect glucose curves.

[Ron Put]

One additional point, on the accuracy of the CGMs: The first 12-24 hours are kind of random, with dubious value in terms of accuracy. This is confirmed by numerous posts and reviews from actual diabetics who compare it to their blood glucose data. It seems to settle down after that.

I am in about 48 hours on the second sensor and this one appears more in line with my lab test, showing an average of about 85. The spikes are also about 10 points lower and so are the dips (it has gone down to the mid-60s on a couple of occasions).

The first sensor appeared to track about 10 points higher than the second one and the pattern is fairly consistent so far.

 

mccoy, did you go through only a single sensor? Based on my reading and experience, while the curve pattern may be valid, the actual values may be a bit off depending on the sensor.

[mccoy]

1 hour ago, Ron Put said:

mccoy, did you go through only a single sensor? Based on my reading and experience, while the curve pattern may be valid, the actual values may be a bit off depending on the sensor.

Ron, I went thru a single sensor, then I installed one on my son and another on my wife. My wife got the worst one in terms of reliability, with many measurements way off the beam. Mine was good the first 10 days, then the baseline drifted upwards. My son refused to get pricked but the signal seemed to be pretty regular.

Apparently, CGMs must always be checked against strip tests or traditional lab tests.

The glucose spikes are just one of the parameters of interest when assessing glucose metabolism. The others are fasting glucose and Hba1c. I can assure you that the low-carb community is well aware of the importance of all of them. Actually, the transients, or spikes, are the less studied of the three, maybe because GCMs are relatively recent. It is believed that a frequent recurrence of high spikes (and slumps) can be deleterious, whereas the ADA only states that after 2 hours from food intake the glucose concentration should be below 140 mg/dL.

Re. optimal diet. I only know that I would have great difficulties in eating high carb low fat. Starches tend to make me full and unsatisfied, I can only eat modest amounts of them. Suffice it to say that I am Italian but I don't eat pasta, I almost hate it.

Edited July 8, 2022 by mccoy