FDA stands against using smartwatches to monitor blood glucose | Hacker News Reader

FDA stands against using smartwatches to monitor blood glucose

9to5mac.com

43 pointsalwillis66 comments

Typical usage of blood glucose monitoring (by diabetics) is simultaneously safety critical, and only borderline good enough accuracy. There's a standard metric for accuracy here, MARD (mean absolute relative difference).

There is a long history of companies promoting themselves to investors as having working noninvasive glucose monitoring technology, which, if you look at the actual accuracy scores, is only moderately correlated with glucose, nowhere close to the accuracy required. If Apple ships a noninvasive glucose monitor with an MARD of 20, the primary effect will be to set a trap for the stupid: diabetics will attempt to use it to save money, since it will be much cheaper than using real CGMs (which are consumable), and people who try to do this will die.

Alternatively, you allow this technology to be sold as a clearly labeled "not accurate enough for treatment of diabetes", and the profit allows more investment in the tech so that one day everyone can benefit from accurate, non-invasive blood glucose monitoring.

Or perhaps we should ban everything that can enable "the stupid" to harm themselves.

There’s far more options than just those two. It may not be physically possible to have non invasive glucose monitoring of sufficient accuracy, until it’s actually demonstrated we just don’t know.

The general point of the FDA is to prevent snake oil salesmen from conning people which makes this decision in line with their core mission.

> It may not be physically possible to have non invasive glucose monitoring of sufficient accuracy.

That's a pretty bold statement. Not economical with current technology? Sure. Not physically possible, at all? I strongly doubt it, and I would love to know the reason for thinking it.

The “may” makes it not very bold.

The standard meaning of "non-invasive" with respect to blood glucose monitoring is that you measure blood glucose without the need to puncture/penetrate the skin.

We know how to accurately test the blood for various things if we can actually obtain a sample of the blood itself. This is straightforward, but you have to puncture the skin to get the blood.

We are starting to be able to infer/guess the blood content of other things, e.g. oxygen, with "useful" accuracy, by measuring externally-detectable changes, such as the color of blood vessels when you shine lights through the skin.

But nobody has ever (so far) been able to do this with glucose. It may be possible one day, but we don't know that it is. Even with sensors 1000x or 10000x more precise than we have today, the glucose content of blood still might be just be one of those things that just doesn't manifest enough external changes to be detected non-invasively. Thus, we still might not be able to do it 100 years from now.

(I hope we can; I know people with diabetes and I also just want my watch to give me this data about myself because, I mean, why not?)

My old dog could sense my mother's low blood glucose from across the room, so I'm assuming we'll all be breathing on our watches before too long.

But was it sensing low blood sugar itself or the behavioral changes of low blood sugar?

She'd wake her up in the middle of the night, but there certainly could have been some behavioral cue.

It was sensing her keto acidosis caused by low blood sugar. It has a distinct fruity smell which any dog can be trained to recognize.

I remember seeing some results from a group using OCT imaging[0] that was on-par with a CGM, just bulky and requiring multiple high-end GPUs to actually run. So it's in principle possible, if nowhere near practical currently

[0]: pretty sure it was this https://pubs.acs.org/doi/abs/10.1021/acsnano.1c00497

I presume you’re thinking of something else?

That approach seems to require injecting nanoparticles and was tested in a simulation of skin not actual skin. Getting bio monitoring tattoo’s is a promising approach, but it’s more involved than just selling a watch.

That's the Standard Libertarian Reply, and in some cases it's the one I stand with, but... no, not for this. A label which everybody knows will be ignored (when was the last time you read the EULAs of your software start-to-finish) by people who then inevitably go on to die because their smartwatch gave them a wrong number should not be allowed.

> Or perhaps we should ban everything that can enable "the stupid" to harm themselves.

Once again this is a case of people holding absolutist principles when there is no need, or benefit, to do so. We can and should use rational cost-benefit analysis to decide what to allow and what to forbid. What's the benefit of allowing non-diabetics access to inaccurate blood sugar measurements? Minimal: a couple people have mentioned speculative uses about mood or energy level, but that's about it. What's the harm? Colossal: it's a near-certainty that many people will die. This is a no-brainer.

Note that by contrast, the calculus for allowing somewhat-inaccurate EKG measurements on the Apple Watch is in favor of allowing it: nobody is going to live or die on an inaccurate heartbeat trace from their watch, as they can for a blood glucose measurement, so, sure, why not. Again: black-and-white principles is not the correct mode of analysis here.

Well, you are describing the maximally absurd end of that spectrum to make your point, but in the US the FDA does indeed ban all kinds of stuff to prevent both its sale, and use (i.e. to stop "the stupid" from using it).

Blood glucose monitors are sold without a prescription in many countries, and you can buy them with varying degrees of accuracy and use them for whatever you want; not in the US, though, because they are regulated as medical devices.

The same is true for lots of other stuff that could be used in non-medical or less-dangerous medical contexts, like nebulizers.

You can make a reasonable argument for your position, regarding this specific device or more generally. But the FDA has consistently rejected it.

Worth noting that glucose meters are OTC. Continuous monitoring solutions generally seem to require a prescription, though there are telehealth companies that will help you sort that out. According to the article what the FDA really objects to is “non-invasive” glucose monitoring devices of any sort, on the grounds of general lack of evidence of accuracy.

Yeah, noteworthy point.

Also, I think the article is weirdly sensationalized — the FDA isn't "against" anything here. If you click through to the FDA's bulletin[1], it's merely stating for clarity and public awareness that there are currently no non-invasive CGM devices that actually meet existing standards for accuracy. If somebody did produce one, it could be validated and approved via the same processes used by all the existing CGM products which have been approved.

[1]: https://www.fda.gov/medical-devices/safety-communications/do...

> Or perhaps we should ban everything that can enable "the stupid" to harm themselves.

Quite a bit of capitalism is predicated on the market not being very bright.

The agency that approves medical devices warns people with serious medical conditions that using devices that haven't passed their standards is a risk. What a shocker.

As the article says, tech companies have been able to get FDA approval for the ECGs on smart watches. Tech companies should do the same for any other medical features they want to sell.

I am glad to have stringent standards that are upheld and enforced.

Approved for what? There is a difference between approved for people with and people without diabetes. If you have diabetes you may want this, but if it isn't very accurate it isn't going to do what you want. If you don't have diabetes though this could be a useful thing sitting on your watch, ready to give an alarm that you need to see a doctor because things are looking bad - people often go diagnosed with mild diabetes for a long time (depending on when their next physical is) and starting treatment for those people sooner can improve long term quality of life. Or if you do have diabetes maybe this can sometimes say do an extra blood sugar test at a time when normally you wouldn't (so long as there are not too many false positives)

It would be useful and probably safe for a diabetic that is not using insulin. Even pre diabetics, just understand how your blood sugar reacts to various foods and activities is really helpful for people trying to manage the disease.

If it's not accurate, it's not going to be helpful.

I'm a paramedic, and for both IDDM and NIDDM patients, watching those numbers, etc., and experimenting, or "chasing the numbers" often leads to more problems than not.

The issue for those patients is that hypoglycemic states are acutely problematic in a short time frame AND lead to decreased alertness, impaired judgment and decision making that can make all these things easily cascade.

And when the numbers are off? It's very dicy. It's not just akin to monitoring your life metrics.

For everyone else, DM = diabetes mellitus and ID and NID = insulin dependent and non insulin dependent, as best as I can tell.

Insulin dependent patients could be considered to have severe diabetes or type 1 (autoimmune/“kid” diabetes).

In a nut shell.

I think even something that is directionally correct could be useful for “pre diabetes “ or non diabetics. A little color coded read out instead of numbers and a general trend may help to keep the numbers at the forefront of your mind. IMO, the susceptible population that I try so hard to get them to start thinking about changing lifestyle just tend to not think about how it is affecting their health and they prefer to wait until they have “full on diabetes” requiring medication. It’s like watching a car crash in slow motion, maybe a green, yellow, red color gradient would help.

You make a good point here, and yes IDDM -insulin-dependent (Type 1), NIDDM (Type 2), as you say (although for clarity - it is fairly uncommon, but possible to get adult-onset IDDM).

If you were wanting to capture trends in pre-diabetics and maybe are looking at ranges, as you said, that could be a useful tool, definitely. "Lifestyle, not (directly) health". Hah, next step, watches that take your A1C blood labs (simplifying, but often used to get a good picture of average BGL over the last three months or so).

The A1C helps, but there are a lot of people who for whatever reason are just not great at taking care of themselves. They don’t do any more reading/research after the Dr explains it to them, they won’t/can’t track carbs and won’t use test stops. They are really disconnected from how their body reacts to food.

Tech companies are doing that. It's legacy health companies that have been trying to sell glucose monitoring watches (Masimo, etc.). Apple is not marketing their device as that because their tech is not yet ready.

As usual you can count on modern tech companies to be better for everyone. Good for Apple.

What??

Apple fans are still bitter at Masimo, evidently.

Masimo doesn't sell a glucose monitoring watch[1]. And is not trying to. The continuous health devices they do sell ARE FDA-cleared, because Masimo makes 99%+ of their revenue from selling accurate medical devices. That would be, to use a crude analogy, dangling your private parts within snapping distance of a crocodile.

Your whole statement is just inaccurate and derogatory.

The W1 watch doesn't do glucometry. It does heart rate and oximetry, with FDA clearance. The upcoming Freedom watch will also do hydration, pulse rate variability, respiration. And will do all of those with FDA clearance. And not glucometry.

Your whole stance is not grounded in reality.

If you want to carry water for a predatory corporation so be it. There's a reason your comment sounds like an ad.

Hah. I have an Apple Watch, basically everything Appple has released except the AVP. Pro Display, cheesegrater Pro, Studio. And you can verify that from my comment history way way back.

But that doesn't mean Apple wasn't predatory here. What they did with Masimo was straight up "brain-rape" (from Silicon Valley). The $7B Masimo was "predatory" to poor $3T Apple? Hah, laughable.

And to top it off, I provided factual rebuttals to your inaccurate and false assertions. So you come back with what? An ad hom that basically accuses me of being astro-turf for "legacy healthcare"?

Fuck outta here.

If you can't be civil there's no reason to respond. Don't swear at me.

> As the article says, tech companies have been able to get FDA approval for the ECGs on smart watches. Tech companies should do the same for any other medical features they want to sell.

The issue here is that the FDA has doubts that any of these devices, as they are currently, and also with some significant physiological challenges (think Theranos "this may just not be possible"), are able to do NIBGL accurately, not that "hey, just make sure you get approval, okay?"

(NIBGL - non-invasive blood glucose level)

That's just 9to5Mac's editorializing. The actual FDA article [1] reads to me like "guys' heads up that none of these are approved". I see nothing to suggest the FDA means "we will never approve one" or "we doubt the underlying technology can ever work".

[1] https://www.fda.gov/medical-devices/safety-communications/do...

The title is more alarming than the substance of the article. Apple is investing in a novel technology that may be able to measure technology using light waves, but the article says it's still in early stages and there's no rush to market with it.

Given Apple's history with ECG and other Apple Watch, FDA approved tech there's no nefarious scheme to deliver anything that doesn't pass approvals.

Honestly, the author of the article is purposefully creating controversy where there is none or ignorant of how the FDA works. The FDA isn't "against such technology" -- they just haven't approved any yet.

Just my two cents, but as a type 1 diabetic: I either want very accurate blood sugar readings, or none at all. And from what I know about people who don't suffer from diabetes: there is almost no reason to try and find out what your blood sugar is doing, because your body regulates it so well.

So I just don't see the point of this. Non-invasive glucose monitoring just doesn't seem feasible, the accuracy doesn't help diabetics and it certainly doesn't help non-diabetics.

I'm a naysayer on this, but I do hate having to continuously test my sugar and buy expensive CGM gear. Please prove me wrong?

I’m not diabetic, but I can have intense energy/mood swings likely related to blood sugar. Non-invasive monitoring does sound nice to have a better understanding of what’s going on.

If I were in your situation (which granted I am not) and mood swings related to blood sugar was a concern, I would pick up a simple finger-prick blood testing kit. It is fast, safe and most of all very accurate. That would then either put my mind at ease or convince me to see a doctor.

Again, having what might be very inaccurate measurements of blood sugar won't help in this case either, I fear.

I'd just go to the doctor.

I don’t understand this view point at all. Widely deployed tech that is not as accurate has a chance to uncover trends no one was monitoring for and to improve over time as they have more data and iterate on the tech.

If we took your approach with literally anything in the tech world we’d have nothing since the v1 wasn’t perfect.

How does it hurt you at all if people have access to a rough estimate of their blood sugar levels?

I'm glad they are holding their ground.

Also, I remember reading "the glucose revolution" and I found it interesting that blood glucose is only part of the picture. We currently don't measure fructose level, and that might have effects separate from blood glucose levels.

Hopefully folks will go back to the drawing board and either lower the cost of the other continuous measuring devices, or move current tech onwards to acceptable accuracy.

>We currently don't measure fructose level

Apologies, but I'm unsure if you're saying that no devices measure fructose level, or just the watch doesn't measure fructose level.

as far as I know no continuous measuring device can measure your blood fructose level. It was mentioned in the book, saying that different kinds of foods might spike your blood glucose level in a similar fashion, yet behind the scenes blood fructose could be wildly different, with a different body reaction.

For anyone interested, other useful measurement are a fasting glucose and fasting insulin to calculate a homa-ir (insulin resistance). I personally think everyone , non diabetc, should get one. Not that expensive to buy it yourself at a direct lab consumer purchasing website (USA).

As much as I think something like this would be a huge benefit, we might need more regulation for these bio-monitoring devices that all seem pretty unreliable in aggregate. In my experience sleep trackers, body fat % scales, calorie trackers, even heart rate meters are 'okay' but have all given ratings that were so bad that I just can trust them. A tracked yes or no question like, "Do I feel like I got good sleep last night?" seems way better than a sleeptracker that can basically only be relied on to tell if I'm in the bed or not.

There should be some requirement that you study and publish the correlation of your device to the current medical gold standard for a reasonable sized population. So a sleep-lab sleep score vs a mattress or fitbit tracked sleep score. Or a body-scale fat% compared to a hydrostatic weighing.

Blood glucose is hard. Witness Theranos' attempt to do blook analysis without sufficient samples.

It apparently can be done from surface contact, with a special patch that measure interstitial epidermal fluid. Do these watches have that?

No, they want to use light refraction to measure the glucose level.

"with a special patch that measure interstitial epidermal fluid."

Yes. Dexcom and Senseonics (Eversense)

"It apparently can be done from surface contact"

It can also be done without pircing the skin. https://quantune.com/

But more complex than Apple thought.

What's the TL;DR on that Quantune device? Does it achieve a new level of noninvasive accuracy?

(I did visit the website but there is no information (possibly because I'm in Japan which has regulations around public disclosure of information about medical devices that haven't been approved — which is enough of a hassle that sometimes companies just basically show a blank page).)

"What's the TL;DR on that Quantune device?"

Interesting concept. QCL with photoacoustic detection.

"Does it achieve a new level of noninvasive accuracy?"

I think so.

I believe the correct approach is to create a non-medical label for the test. Our watches have heart-rate monitors, not EKGs. An EKG is a medical device with highly accurate measurement of heart-rate for medical purposes. A heart-rate monitor, gives a decent approximation, not good enough for medical purposes.

Nobody calls an EKG a heart-rate monitor. The correct term would let a consumer have some insight into their blood sugar levels, but a diabetic would clearly not be confused into thinking this was a replacement for the monitoring they require.

My fear is that "a diabetic would clearly not be confused into thinking this was a replacement for the monitoring they require" is an impossible bar to clear. We've seen how good ordinary people are at using technology even when it has clear documentation.

Another post on this article also points out that traditional monitoring costs money on an ongoing basis, which would create a temptation to try and 'get by' with tech like this even if it's not adequate.

These are good points, but we can't ignore that this technology is valuable for non-diabetics. We also know that "preventing" technology from being used is a losing battle - though I'd be happy to hear of examples contrary to this.

If a diabetic is going to attempt to use the device as a diagnostic, but they are aware that it is not fit for that purpose, that is their own choice. Like injecting insulin by "feel".

Perhaps a better approach would be that insulin be delivered in collaboration with a certified device. That would close the loop on the amount of insulin being injected.

I'm no expert on diabetes though, so i don't know the actual implications of that.

Going to buy more Eli Lily stock, that thing will never go down this way.

Keeping people from seeing their real-time blood glucose levels is just a move to keep them from becoming aware of how bad their sugary foods are and their pre-diabetic health.

Fructose is way worse than glucose.

Glucose is still rather bad, but I'm on a fructose limiting diet anyways.

Imagine we could have a near real time effect of what we eat and then we discover that high fructose products are really really bad.

Also imagine that the corporations that produce those products are the worse when it comes to corruption through lobbying.

I don't know what really really bad means to you, but we can look around and see people who have eaten fructose their whole lives to no obvious effect. Almost every person I see.

So it's not going to meet my personal definition of really really bad.

It's already clear and well known that it's bad.

Cancer rates will soon be over 50%, and obesity and heart disease have never been higher.

50% of what? Have never been higher per what, a specific age group?

More than 50% of all people will have cancer in their life.

And that’s due to a magically bad meme ingredient rather than the explosion of calorie dense foods and norms in the food supply?

Average calories consumed peaked in the early 2000s, we eat less calories and exercise more since that time period but obesity rates are up by at least +%20. It's not the calorie counts.

> heart disease have never been higher.

As I understand it, this is false, so I'm going to ask for a citation.

Between the decrease in smoking and the usage of statins, it seems that heart disease has dropped dramatically.

Funny, I look around and see people every day in terrible condition who have been consuming oodles of fructose their whole life.

I don't think we will discover that. Fructose is pretty benign chemically (though it can be inflammatory as fructans). Instead, I think we'll find some kind of food additive or chemical is causing autoimmune issues that create a whole host of secondary effects like hypothyroidism, CFS, and cravings/obesity that potentially link to T2 diabetes.

I don’t think that’s true. People with normal endocrine response (not diabetic or otherwise impaired) don’t see significant movement in blood glucose. It would very likely be “in the noise” as far as the light based sensors go.

Thing is most people don't actually care, and individual measurements are usually just vanity metrics. We already know that fructose is poisonous. We already know that processed foods are engineered to keep you hungry. But people eat this stuff anyway. I no longer use fitness devices, and the people I know who use them are almost always in suboptimal health and make no progress, no matter how many "steps" or "calories" their devices claim they are producing.

Stop eating poison and one's health usually improves – no tech needed. Better yet if you give your body adequate time to dig in to its own fat stores.

People want easy solutions and to feel smart. Smart watches and Oura rings claim exactly that. Don't research, don't think – just hit a reductive number so your watch can tell you you're "on track" and give you permission to eat some Cheetoes and coke.

"Bad" in what sense? Continuous glucose monitors have been available for several years and can show blood glucose spikes (or lack thereof) after eating certain foods. But you're not going to get any useful data about fructose from a wrist device. The issues with fructose seem to be interrelated with sodium and uric acid.

https://peterattiamd.com/rickjohnson2/

Lately endurance athletes have been increasing their fructose intake during races and hard workouts in order to maximize energy availability.

https://www.trainerroad.com/blog/how-to-use-carbs-for-maximu...

Reply on news.ycombinator.com