They trained an ML model that predicts blood glucose based on age, sex, GPS location, and Apple Pay transactions. Just bought a #3 combo at Taco Bell and arrived at home? Blood glucose spikes 15 minutes later.
From the linked article: Apple has been revealed to be the largest customer of the British electronics start-up Rockley Photonics, The Telegraph reports.
Rockley Photonics has developed non-invasive optical sensors for detecting multiple blood-related health metrics, including blood pressure, blood glucose, and blood alcohol levels, many of which are only normally detectable with more invasive dedicated medical equipment. Rockley's sensors beam infrared light through a user's skin, much like the existing sensors on the back of the Apple Watch for detecting heart rate and blood oxygen levels.
It would be very interesting how this would reliably work across a diverse population without some sort of calibration step. I am skeptical, but it is Apple.
This still doesn't explain how it works. Oxygen and heart rate can be done with light, this is well known, but how does it measure glucose and alcohol levels and blood pressure?
If someone is really drunk you can smell the alcohol standing near them. Not sure about glucose but possibly it's detecting chemicals diffusing through the skin or pores?
Blood oxygen levels are also measured optically already. If you can find an optical property of blood (that depends on a particular substance) which can be measured through the skin (infrared light can penetrate skin and flesh to some degree), you can build a sensor for it.
Could be something else. The wikipedia page on this technology (https://en.wikipedia.org/wiki/Noninvasive_glucose_monitor) mentions “bioimpedance spectroscopy, microwave/RF sensing, fluorescence technology, mid-infrared spectroscopy, near infrared spectroscopy, optical coherence tomography, optical polarimetry, raman spectroscopy, reverse iontophoresis, and ultrasound technolog“
Sure but oxygenation is strongly correlated with hemoglobin color, so it's very easy.
Blood pressure, on the other hand, is very difficult to measure, and it fluctuates wildly based on a wide variety of confounding factors. It would be amazing if they figured out a way to get this information optically.
> Rockley’s optical technology enables the miniaturization of sensing devices necessary for the evolution of a wearable spectrometer [0]
If you have a wearable spectrometer (i.e. measure the wavelengths that are reflected), and a variable light source (i.e. control the frequencies sent into the body), you can isolate repeating patterns, even for molecules like glucose: https://youtu.be/xMa1BQ8z9C0?t=541
I'm still not sure how they can tell the parts per million, etc. Probably through multiple samples over time, or multiple veins.
unfortunately, the real, invasive blood glucose meeters are already pretty inaccurate. The current requirement in the US is +-15%, before 2016 it was +-20%.
Not to be a buzz killer but I don’t think that’s ever happening. Displaying an actual number is bound to have kids try to max it. Using the legal limit and a colour code might open it up to liability.
You can already buy devices that display estimates of blood alcohol concentration derived by measuring breath alcohol concentration. I’ve never heard of any of their vendors being successfully sued on the theory you present. I doubt a unit somehow measuring it through the skin would be any different.
I think the issue would arise if someone used it as justification of normal blood alcohol limits and then drove, was stopped and then found to be sufficiently intoxicated to be illegal. Then sue. If apple say cannot be used in this context, then what context is it actually helpful?
That already happens with the existing breath-based devices. People claim they measured under the limit on their own device, then still get caught by law enforcement. Sometimes law enforcement has more accurate devices. Blood alcohol concentration can rise immediately after drinking, so sometimes you can measure under the limit when you start to drive but be over the limit when pulled over. Some people just lie, and claim they measured under on their own device even when they didn’t.
I’ve never heard of a manufacturer being successfully sued over this. If it was going to happen, it probably would have happened by now. The law has never had much sympathy for drunks.
Even if it’s accuracy was not that great, say <25%, 25-75%, and >75% of legal limit would still be useful to know as some places have heavy handed bartenders, were you could end up with more than your expecting if you are counting drinks.
Even having crude data will let me know how fast various formats of alcohol are uptaked and processed out so if I have a drink with friends, I would know ahead of time if I should take transit vs walk in park after and then drive or stick to alcohol-free drinks (most people I know are cross town in a way public transportation is not cheap/quick but a car is).
Edit: I do use a breathalyzer on myself now, but they are annoying to take samples, and are not super accurate either
Where I'm from, we already make a bit of a game out of trying to get a "high score" on the BAC checkers that are at most pubs - and that costs money. I can 100% see this turning into a game for a lot of people (me included, if I'm honest)
This and the parent post are both right on. I would love that BAC thing. It would take me from 0% likely to ever own an Apple Watch, to almost 50%. It would reduce drunk-driving scenarios because people are going to be monitoring and pacing their bar/party sessions.
But I foresee lots of lawsuits making Apple shy to offer it. No matter how smart Apple is at making it legal, there are always amateurish or vindictive governments that will make laws that cause Apple to risk responsibility.
If you mean it can tell you when to drive, I know that it doesn't matter how much alcohol you have in your system, you can always get nailed for a DUI. It might be easier to prosecute with the numerical value being over the legal limit, but even if you're under the limit, the police can still argue you're too drunk to drive.
I feel like this is going to be a new popular pastime here in Australia, we already make a bit of a sport out of blowing into the BAC checkers at the pub I can't imagine how out of hand it's going to get if we've all got BAC checkers strapped to our wrists.
I was never interested in "smart watches" but an affordable and non-invasive blood glucose monitoring would be a game changer for me, even if it is not accurate as the ones diabetics have to use (I am not diabetic) and if it "just works" without fuss like how Apple usually does it.
If I'm understanding the blood glucose monitoring feature correctly, this would be very valuable for me. I don't know that I would want to wear an Apple Watch all the time (my Pebble does what I need on a daily basis), but it would be very helpful to be able to wear one for a month or so to see how my body responds to various foods.
For example, I've heard that some people have spikes after eating bananas, but other people do not. Also, rice can be very bad for some people (and different kinds of rice have different effects). I'd be curious to wear one for a month or so just to see how my body reacts to various foods and combinations of foods. I think it's possible to rent devices that do this, but it's hundreds of dollars for not that long of a trial. If I could spend $400, gather data for a month, and then sell the device in like-new condition for $250, that'd be a win in my book.
Unless I had a medical need, I would not want to wear any monitoring devices 24/7. I think it would contribute more to overall anxiety and stress than do anything helpful.
I know some people who can't even wear a fitbit to bed because it causes anxiety. I am open to the idea of wearing a device that monitors blood glucose if (1) it could accurately measure blood glucose, and (2) the information it provides me passively/via notification can help me to make significant improvements in my health.
For example, if I could make my liver 20% healthier by eating more moderate portions of bread in a meal where I plan to eat dessert, it would be useful to know that.
On the other hand, I don't think I have to wear the device forever to gain this information. I could learn about what makes my blood glucose spike after a month or two, and then just carry that knowledge with me going forward.
Honestly it's not like these stats are always present on the watch face, unless you choose for things like your heart rate to be present. For ECG for example, you have to specifically go into the ECG app and hold your finger on the crown for 30 seconds just to get a scan.
Exactly. I would much rather have these sensors embedded in elsewhere where I can make a conscious decision to activate the sensor as a part of my routine. Something that sits on my body all the time is just to anxiety-inducing.
I guess it depends on how you look at it. I got an apple watch that I wear all the time and I occasionally find the data interesting. For instance, I was able to see in absolute terms just how much cutting out coffee improved my sleep (went from averaging ~6 hours 10 min to ~7 hours asleep). I don't look at the data all the time though.
I hold/held the opinion. I've asked several of my doctor friends what they think of a future like this. They are all resoundingly pessimistic, due to the false positives they think would result and a general sense that it would result in unnecessary interventions.
It made some sense to me, but ultimately considered it would work out for the greater good.
If you think about something like cardiovascular disease or diabetes, it seems incredibly obvious that having a continuous stream of data with blood glucose, pulse, oxygen etc. would be extremely valuable.
So I feel like your doctors maybe don't understand or appreciate technology.
That PDF covers what they can monitor, including glucose indication, continuous glucose, BP, and other biomarkers. Slides 25-28 are mostly about their product roadmap and unnamed existing alpha customers.
I really hope this happens (well the blood glucose thing), but this rumor has come around more than once. Admittedly this is the first time I've seen it name a company that is actually working on the tech. I wonder if including this will require FDA approval in the US as a medical device? And will I be able to get my health insurance to pay for it?
They don't have FDA approval as a medical device for the HRM, yet. They have a lower "clearance"(?) standard, I believe, which won't meet insurance or other requirements.
well, one of the existing CGM (continuous glucose monitor) products on the market, Freestyle Libre, offers phone-based NFC monitoring of their sensor everywhere EXCEPT in the U.S., because the FDA hasn't given approval yet. It's been "real soon now" for almost a year.
Not the sensor itself, mind you. That's been approved forever, just the ability to read it with an app instead of their dedicated NFC reader.
Existing watch heart rate sensors are next to useless IMO even when they work. I've always been skeptical of fitness features on watches. But continuous blood pressure and glucose monitoring would definitely be worth buying and wearing a new watch. Maybe even switching from Android to iOS if it was Apple exclusive.
I'm a little skeptical that it's possible to get accurate absolute measurements of those things purely optically, through skin and hair and while the body is in motion, across all skin types, ages, etc. That would be pretty magical.
I compared a series 4 Apple Watch to my Garmin chest strap HRM and they gave pretty much the same readings. The Apple Watch was well within practical usability for training at the gym.
I can't speak to the quality of the following series, but if that isn't good enough, what is?
BP monitoring would be massively useful. Unless I’m missing something, there isn’t an existing product on the market which measures blood pressure without some kind of inflatable cuff. For people who need to measure BP regularly, this would be quite literally life changing.
The fact that nothing does this makes me wonder how it would be possible apple could be first. Surely if it was possible to do accurately, some other company would have released something.
There are some measures to make it work (such as [0] which uses bioimpedance ), but they are small scale, and I suspect reliability and accuracy will be an issue.
It’s a bit of a circular logic though, isn’t it? Apple can’t be first, therefore they aren’t first and the sensors are useless. Personally, I have no opinion on the technology, but Apple is actually commonly the first company to use some new technologies in mass-market devices.
What they usually do is find a small company with some compelling technology and then either fund it to gain exclusive or privileged access, or buy it outright. The story here (that a small company working on cutting edge sensors has close links with Apple) fits that pattern perfectly.
Apple is rarely ever the absolute first. Usually similar things have existed for a while and apple comes in and makes them much better and more accessible.
Sure, maybe apple is first this time but I have a hard time believing Apple is at the cutting edge of new medical technology.
I believe (could be wrong), that Apple was the first to have a mass-market wearable with a ECG feature. So you don’t even have do look that far to find a refutation.
I don’t know about how innovative these rumored sensors may to comparatively, though.
It would be amazing if they could do continuous BP monitoring throughout the day, without user having to trigger it. I've always struggled with tensing up when taking blood pressure, raising my reading.
I can't see how this is possible. There is no published research on using light to detect blood glucose and blood alcohol with any accuracy. Blood pressure is just about possible [1] if you calibrate with a real pressure sensor but even then the results aren't great.
I relay the words of my housemate that's a nurse. How are they able to do this on simply your wrist when they use large machines with multiple sensors in hospitals?
My argument is that those machines are highly accurate, or at least the most accurate in order to diagnose.
This extends to the other measurements too like respiratory rate and bloody oxygen saturation.
73 comments
[ 3.6 ms ] story [ 133 ms ] threadIt somehow can see microscopic features in moving blood through the skin and blood vessel walls?
Not sure how close it is to being out of the lab but it definitely seems doable.
Rockley Photonics has developed non-invasive optical sensors for detecting multiple blood-related health metrics, including blood pressure, blood glucose, and blood alcohol levels, many of which are only normally detectable with more invasive dedicated medical equipment. Rockley's sensors beam infrared light through a user's skin, much like the existing sensors on the back of the Apple Watch for detecting heart rate and blood oxygen levels.
Edit: formatting
https://en.wikipedia.org/wiki/Near-infrared_spectroscopy
That page also claims there isn’t any clinically or commercially viable product yet. That seems out of date to me, given that https://www.labiotech.eu/best-biotech/needle-free-glucose-mo... mentions 8 devices you can buy now.
"needle free" !- "non-invasive".
Several of those devices are implanted subdermally.
And now I can’t stop thinking about wearing a Raman spectrometer on my wrist :)
Blood pressure, on the other hand, is very difficult to measure, and it fluctuates wildly based on a wide variety of confounding factors. It would be amazing if they figured out a way to get this information optically.
https://rockleyphotonics.com/wp-content/uploads/2021/03/Rock...
https://patents.justia.com/assignee/rockley-photonics-limite...
If you have a wearable spectrometer (i.e. measure the wavelengths that are reflected), and a variable light source (i.e. control the frequencies sent into the body), you can isolate repeating patterns, even for molecules like glucose: https://youtu.be/xMa1BQ8z9C0?t=541
I'm still not sure how they can tell the parts per million, etc. Probably through multiple samples over time, or multiple veins.
[0] Page 18: https://rockleyphotonics.com/wp-content/uploads/2021/03/Rock...
And the killing of buzz is bi-directional too!
I’ve never heard of a manufacturer being successfully sued over this. If it was going to happen, it probably would have happened by now. The law has never had much sympathy for drunks.
Even having crude data will let me know how fast various formats of alcohol are uptaked and processed out so if I have a drink with friends, I would know ahead of time if I should take transit vs walk in park after and then drive or stick to alcohol-free drinks (most people I know are cross town in a way public transportation is not cheap/quick but a car is).
Edit: I do use a breathalyzer on myself now, but they are annoying to take samples, and are not super accurate either
But I foresee lots of lawsuits making Apple shy to offer it. No matter how smart Apple is at making it legal, there are always amateurish or vindictive governments that will make laws that cause Apple to risk responsibility.
For example, I've heard that some people have spikes after eating bananas, but other people do not. Also, rice can be very bad for some people (and different kinds of rice have different effects). I'd be curious to wear one for a month or so just to see how my body reacts to various foods and combinations of foods. I think it's possible to rent devices that do this, but it's hundreds of dollars for not that long of a trial. If I could spend $400, gather data for a month, and then sell the device in like-new condition for $250, that'd be a win in my book.
For example, if I could make my liver 20% healthier by eating more moderate portions of bread in a meal where I plan to eat dessert, it would be useful to know that.
On the other hand, I don't think I have to wear the device forever to gain this information. I could learn about what makes my blood glucose spike after a month or two, and then just carry that knowledge with me going forward.
I mean, the potential health benefits in terms of preventive care seem to be incredibly large to me.
It made some sense to me, but ultimately considered it would work out for the greater good.
So I feel like your doctors maybe don't understand or appreciate technology.
As part of merging into a SPAC, Rockley announced Apple as its largest customer (https://www.telegraph.co.uk/technology/2021/05/01/apple-watc...).
That PDF covers what they can monitor, including glucose indication, continuous glucose, BP, and other biomarkers. Slides 25-28 are mostly about their product roadmap and unnamed existing alpha customers.
(I don’t have a position in or relationship with this company. The PDF is linked to from https://rockleyphotonics.com/investors/)
...I'd stay the hell away if I were you.
Not the sensor itself, mind you. That's been approved forever, just the ability to read it with an app instead of their dedicated NFC reader.
I'm a little skeptical that it's possible to get accurate absolute measurements of those things purely optically, through skin and hair and while the body is in motion, across all skin types, ages, etc. That would be pretty magical.
I can't speak to the quality of the following series, but if that isn't good enough, what is?
[0] https://pubmed.ncbi.nlm.nih.gov/31312808/
What they usually do is find a small company with some compelling technology and then either fund it to gain exclusive or privileged access, or buy it outright. The story here (that a small company working on cutting edge sensors has close links with Apple) fits that pattern perfectly.
Sure, maybe apple is first this time but I have a hard time believing Apple is at the cutting edge of new medical technology.
I don’t know about how innovative these rumored sensors may to comparatively, though.
And could such hacking lead to injuries or deaths?
[1] https://aktiia.com
My argument is that those machines are highly accurate, or at least the most accurate in order to diagnose.
This extends to the other measurements too like respiratory rate and bloody oxygen saturation.