EEVblog Electronics Community Forum
General => General Technical Chat => Topic started by: frogg on January 16, 2020, 01:18:54 am
-
Of all the devices that haven't been commercialized (or really developed at all) yet, accurate non-invasive blood glucose monitoring is pretty important. Unlike continuous heart-rate monitoring and blood oxygen sensors which are now commonplace, and high resolution ultrasound which has become so portable and simple that it has become a subscription-based phone dongle and app, there's no real existing device that does the job. Until we build one, it's obviously impossible to make them miniature, portable, or cheap.
What's the most promising leap forward currently for non-invasive blood glucose monitoring? What is the technology that hasn't been invented or isn't well-enough understood that keeps a viable commercial product from emerging?
-
Unlike oximeters, which use difference in absorption spectrum of red cells with and without oxygen, or heart rate monitors, which rely on signals detectable on the skin surface, blood glucose level measurement requires acquiring a sample of blood. So far I haven’t heard about any indicators of glucose levels that are detectable on the surface or by passing some kind of radiation through tissues. The best I’ve seen are those coin-sized devices glued to the skin, which periodically prick you with a tiny needle to acquire a sample of extracellural fluid. But that method, while allowing automated and frequent sampling, is imprecise and is not measuring the instantenous value of blood sugar: the time needed for extracellurar fluid to contain as much glucose as blood makes it act as a low-pass filter, averaging the output over a quarter hour.
I suspect that the best thing would be not non-invasive monitoring, but a tiny, minimally invasive implant. And for that the main obstacle is, as it is common nowadays, a power source.
-
Theranos is well known attempt to do this. There is even book or documentary about CEO Elizabeth Holmes.
(https://upload.wikimedia.org/wikipedia/commons/b/b3/Elizabeth_Holmes_2014_%28cropped%29.jpg)
-
The power source for an implanted sensor shouldn't be an issue, it's not difficult to inductively couple a small amount of power a short distance like through the skin. Service life of a sensor might be a challenge though.
-
I'll show my ignorance of the intersection of chemistry and electronics here. Is there even a good way for an implantable sensor to detect glucose concentration without using some sort of consumable reagent?
The sensor would have to be biocompatible, and able to continue working for a relatively long time.
I can appreciate the need. The market must be huge. But that doesn't mean the solution is easy or even possible.
-
Supposedly the tech for saliva testing is already out there, although there is 100x less glucose than in blood. Diabetics are pay a whopping $1 per test strip.
Google hyped it with their contact lens, (https://en.wikipedia.org/wiki/Google_Contact_Lens) another load of bull as the project was shelved after 4 years. Anyone out there doubt a sensor+wireless IC+antenna+battery can be in your eye? :palm:
https://glucosebiosensor.com/saliva-glucose/ (https://glucosebiosensor.com/saliva-glucose/)
https://hitconsultant.net/2019/02/26/saliva-based-glucose-test-diabetes/ (https://hitconsultant.net/2019/02/26/saliva-based-glucose-test-diabetes/)
-
The only one I knew was the Google contact lens but that went nowhere.
You need to sample blood to measure glucose so almost by definition it will be invasive.
Google hyped it with their contact lens, (https://en.wikipedia.org/wiki/Google_Contact_Lens) another load of bull as the project was shelved after 4 years. Anyone out there doubt a sensor+wireless IC+antenna+battery can be in your eye? :palm:
So I actually did an nterview with Google for a similar role and they told me (this is public information so I'm not sharing anything confidential here) that it was an RFID powered device. I.E. you put your phone up to it and that would power the circuitry within the device and a measurement would be taken. It would certainly be a complex device to manufacture, but I think in 20 years time that kind of technology could be commonplace.
Now someone is going to tell me 1W+ of RFID power next to a dense clump of optic nerve is a bad idea!
-
Now someone is going to tell me 1W+ of RFID power next to a dense clump of optic nerve is a bad idea!
Sometimes the cure is worse than the disease, at least before you realise that you can sell it as a revolutionary solution... In one case you go blind because of diabetes and insufficient glucose monitoring, in the other case you go blind because of diabetes and excessive monitoring. :horse:
-
I'll show my ignorance of the intersection of chemistry and electronics here. Is there even a good way for an implantable sensor to detect glucose concentration without using some sort of consumable reagent?
The sensor would have to be biocompatible, and able to continue working for a relatively long time.
I can appreciate the need. The market must be huge. But that doesn't mean the solution is easy or even possible.
(https://images.squarespace-cdn.com/content/v1/52276cf1e4b0a9346b988071/1510240277916-MVQU5SOLXUC741D6Q3FZ/ke17ZwdGBToddI8pDm48kPKoAxQt6eC7sVWD4onoMXpZw-zPPgdn4jUwVcJE1ZvWQUxwkmyExglNqGp0IvTJZamWLI2zvYWH8K3-s_4yszcp2ryTI0HqTOaaUohrI8PIwq_UFVI7UOd0VoB4kCNQyumNMQ_BOg8RNH_Pbfqis98KMshLAGzx4R3EDFOm1kBS/Illustrasjon_gel_tech.jpg?format=750w)
https://www.glucoset.com/technology-detailed (https://www.glucoset.com/technology-detailed)
I used to work at an ancestor to this company. The sensor fiber is 125um in diameter. Not sure about the lifespan for it. I think sensors in a bloodstream have a tendency to clog up.
The fiber-optic system for reading the sensor wasn't very portable when I worked on it. Pretty power hungry too. That can be fixed, I'm sure.
We aimed for continuous glucose monitoring during surgery, at a hospital.
-
I think sensors in a bloodstream have a tendency to clog up.
I would think so as well, so this approach is fine for short-term continuous monitoring, but not for long term implantation.
The blood vessel would probably also end up making scar tissue around the probe.
And that said, any implanted device, however small and easy to implant it is, is an invasive device.
-
The power source for an implanted sensor shouldn't be an issue, it's not difficult to inductively couple a small amount of power a short distance like through the skin.
Good point. I’ve focused too much on making it different than currently used CGMs, forgetting that the original issue was invasiveness and not having something attached to your body. |O
-
Invasive:
https://diatribe.org/flash-glucose-monitoring
https://link.springer.com/article/10.1007/s13300-016-0223-6
Noninvasive:
https://pubs.acs.org/doi/abs/10.1021/acs.accounts.6b00472
https://ieeexplore.ieee.org/abstract/document/8058721
https://link.springer.com/article/10.1007/s11082-016-0490-5
https://iopscience.iop.org/article/10.1088/1612-202X/aa58c0/pdf
https://advances.sciencemag.org/content/3/12/e1701629.short
-
Might be a market for a AI system monitoring the symptoms which result from improper blood glucose. That answer is likely not timely enough to be desirable for insulin dosing for best health, but it might make a decent backup to prevent really severe problems when the primary method has been compromised because of user error, lack of supplies or any number of other issues. And who knows. As the database grows it might move forward in the diagnostic chain.
I don't know if monitoring the various relatively easily obtained variables (temperature, pulse rate, BP, skin conductivity, O2 concentration, respiration rate, blink rate and so on) can give any useful indication, but clearly they can detect a severe problem so there is at least some chance.
-
Although diabetes management is the most obvious application for non-invasive blood glucose monitoring, I think a real continuous non-invasive device (which could ultimately be scaled down to a compact device) would completely transform how people interact with food and provide an added dimension to their understanding of activity.
Imagine a wristwatch that could tell you when you've eaten too much refined sugar, or too much food. A watch that could tell you to eat a snack, and tell you when you've eaten enough. Just imagine you're downing a can of sugary beverage, and the watch starts beeping an alarm saying that your blood glucose level has shot up as a reminder to think twice about drinking a second can right away.
Or perhaps you know that you will be unable to eat for a 10 hour shift, and you know that you need to eat enough food to cover your energy needs for that period of time. How much do you have to eat prior to the shift? Or perhaps before heavy exercise?
Just like measuring steps has become commonplace, a person could totally map their body's blood sugar levels over time and create a baseline for what is "normal" and "abnormal" for their bodies. Measured every day, 24 hours a day, 365 days a year, I imagine that the meaningfulness of that data would really revolutionize people's relationships with their own bodies (maybe that sounds bad :))
Just a thought...
Thanks for the references and thoughts...good reading!
-
This is a problem that people have been trying to tackle for decades I believe a "John L Smith" regularly updates a white paper on the subject, along with a list of failed companies. Part of the problem with using reflection/absorption of various types of radiation (RF, light, IR etc..) is that you're looking for a chemical signature in a system where many many molecules have the same signature. That and the fact there's only really ~4g of glucose in all of the blood of an average person.
The latest is using AI (well, machine learning) on ECG results, but this doesn't really provide any kind of accurate level of blood sugar, rather warnings about hypoglycemia - the signs of which I'm sure diabetics know about anyway.
I'm sure you could find some companies that claim to have non-invasive glucose measurement (we're not talking about implants either) but I almost guarantee they don't work as well as described. Many just seem to list all of the benefits to such a system, like that is somehow a new idea, but haven't found a way to measure it. Plenty of false claims about.
Edit: Found it: https://www.researchgate.net/publication/317267760_The_Pursuit_of_Noninvasive_Glucose_5th_Edition (https://www.researchgate.net/publication/317267760_The_Pursuit_of_Noninvasive_Glucose_5th_Edition)
-
As google and the others learned, blood sugar is in the blood and nowhere else, some cells picking it up and converting it slowly, other fluids responding rather slow. And slow is what you actually don't want because its already here. This leads to the conclusion that you need something to suck in blood fresh from your vessels. Also, it shouldnt in any way harm you, which means releasing no toxic stuff above what your body can tolerate.
To make it even harder, blood is not nice to stuff inside the vessels, lots of enzymes and the immune system will do its best to get you out. You might trigger that blood clots on the surface of your device, tata, a membrane, separating you and the glucose even further. And the clots cause big trouble when they get washed away and block some vessels eventually.
There is no clean-glucose reservoir, so you get a mix of several interesting chemicals so you need a sensing system that is mostly sensitive to glucose and has low-cross-sensitivity. Mind if we put you to a mass spectrometer-gas chromatograph separating your consitutients?
The easiest way is letting a very selective enzyme (glucose oxidase) do its work and look at the charges exchanged. These enzymes get used up and dont live forever, especially when you more or less cook them off during storage close to or inside your body, so refill required.
There is a bit of hope. you can twist and shake it by electromagnetic radiation, so maybe only a fiber or electrode needs to be in the vessels, leading to, say, a VNA or optical spectrometers.
BR
Hendrik
-
Some of the RF approaches seem rather dangerous because they lack physics modeling of why they show correlation.
Just showing good correlation isn't enough, you don't know the failure modes if you don't know why it sometimes works ... even if that sometimes is every time during your experiments.