There are three basic views when we talk about a smartwatch with a built-in glucose monitor.
Some people say they don’t care about glucose monitoring on a smartwatch since that’s the problem of diabetics. Some think that such technology is fundamentally impossible. Others are ready to pay whatever it takes to get such a smartwatch.
Plenty of fitness trackers accurately measure blood oxygen level (SpO2) and heart rate variability. There are smartwatches with ECG and even cuffless blood pressure monitors. But what’s wrong with blood sugar measurement?
Despite the enormous demand, there is still no glucose monitor that can accurately measure blood sugar levels without puncturing the skin, but why? And does a healthy person even have to concern himself with this?
Let me try to take all of these questions!
Diabetes for dummies
Not all people on earth have diabetes, obviously. According to the latest data1, more than 500 million adults (1 in 10 worldwide) live with diabetes. Half the people with type 2 diabetes don’t even know they have the condition because they have no symptoms.
It was predicted in 2009 that there would be 438 million people with diabetes in 20 years. But we have already exceeded this figure by 100 million people and now expect 643 million patients (not including children) until 2030. Last year one death every 5 seconds was a consequence of the complication of diabetes.
And that’s just the tip of the iceberg. High blood sugar levels slowly but surely destroy people from the inside, causing various diseases seemingly unrelated to sugar.
What is diabetes, and what does sugar have to do with it?
To live, grow, and sustain proper body functioning, we need not only “building material” (that is, protein consumed with food) but also energy.
And it just so happens that the primary energy source for living organisms is sugar or a molecule called glucose (the thing we call sugar and sweeten hot drinks with is a more complex molecule consisting of two simple ones – glucose and fructose).
The brain alone consumes up to 120 grams2 of glucose daily (about 420 kcal), which we get from food. In addition, there is glucose storage inside our body in the form of glycogen molecules, which, if necessary, can be broken down to glucose very quickly and easily.
But here’s the thing about glucose absorption: these molecules can’t get through cell membranes on their own and need the help of transport or carrier proteins called GLUT, which can carry them across the cell membrane.
There are about dozen different GLUT transport proteins: GLUT3 in brain cells (neurons), GLUT1 in erythrocytes, GLUT4 in muscle cells, etc.
Take a brain cell, for instance. Glucose molecules can get into neurons only using the GLUT3 protein. When glucose comes close to a neuron, it immediately passes through the cell membrane due to this transporter:
But something interesting happens when glucose needs to enter cells that use the GLUT4 transporter. The glucose molecule can’t get inside the cell because the GLUT4 is not on the cell surface:
To deliver glucose into the cell, GLUT4 must be brought to the surface of the cell membrane. And the only way to do this is to use a “key” called insulin. This hormone binds with a special cell receptor, and after a series of biochemical reactions, GLUT3 comes to the surface.
Now glucose can easily penetrate the muscle cells through the GLUT4 protein.
But don’t think that other cells (i.e., neurons) will work with glucose properly without insulin. Even though other transporters (GLUTs) pass glucose without using insulin, this hormone is required to break glucose down inside the cell.
So, let’s recap:
- Every cell needs energy
- The source of energy is glucose
- The glucose is useless without insulin
If the pancreas produces little or no insulin, such pathology is called type 1 diabetes. And type 2 diabetes is when insulin is produced but used inefficiently (cell sensitivity to insulin is impaired).
In the former case, it’s vital to monitor blood glucose levels to maintain glucose and insulin balance. Otherwise, you can slip into a diabetic coma.
Type 2 diabetes is dangerous since many people don’t even know they have it. But the constant high blood glucose level eventually leads to complications of heart, kidney, vision, etc.
That’s why people who think they don’t need a smartwatch with a glucose monitor are very wrong. Measuring blood sugar is more critical than currently popular blood oxygen saturation level (SpO2) measurement. And wouldn’t it be great if your fitness tracker could cope with this task?
How is blood sugar measured today?
Back in the day, everything was quite simple. The only way to measure blood sugar was to taste your urine.
As we already know, our cells can’t use glucose without insulin. So if the body can’t get the glucose, it increases the amount of these molecules by sending the pancreas command to synthesize glucagon. This hormone, in turn, causes the liver to break down previously-stored glycogen molecules, thereby extracting glucose.
Glucose concentration increases even more, but cells still don’t receive it. Excess glucose is excreted in the urine, giving it a sweet taste. Hence the name of the disease Diabetes Mellitus (from Latin sweet as honey).
We don’t taste urine to identify diseases nowadays. Instead, we have a particular device called a glucose monitor (more specifically, electrochemical glucose monitor):
And here is how it works. You puncture the fingertip with a lance, then apply a drop of blood to a strip. Blood glucose reacts with an electrode coated with a special enzyme that oxidizes glucose, producing an electric current.
The total charge is proportional to the amount of blood glucose. The device reads this charge and gives you the number of blood glucose molecules.
Normally, the fasting sugar level should be 3.3 to 5.5 mmol/l, and after eating it should not exceed 8 mmol/l.
And don’t be confused by the value mmol/l. It’s just a convenient way of recording the total number of molecules.
In order not to operate with huge numbers (billions of trillions), people decided to count molecules with “packets” called millimole (mmol), where one mmol is equal to 600 million trillion molecules3.
In the US, we use mg/dl (milligrams of glucose per 100 ml or 1 deciliter of blood) instead of millimoles per liter. Accordingly, the range of 3.3-5.5 mmol/l equals 60-100 mg/dl (we just multiply the value in millimoles by 18).
What’s wrong with electrochemical glucose monitors?
It would seem that the problem is solved! After all, there are convenient, affordable, and pretty reliable devices (a glucose monitor should measure glucose with an error4 of no more than 0.83 mmol/l or 15 mg/dl). But it’s not that simple.
First, a healthy person is unlikely to buy a glucose monitor since it’s not the cheapest thing, and there is no need to measure glucose daily. Also, you always have to use disposable test strips (at least 3-4 strips a day), which is not very convenient either, let alone the risk of infection while puncturing the skin.
But even with all that in mind, glucose monitors don’t completely solve the problem as they only make one-time measurements. Sugar levels can fluctuate significantly during the day between each measurement, thereby harming the body.
People with diabetes mostly feel a critical drop/increase in sugar levels (hypoglycemia or hyperglycemia). Still, they do not feel a non-lethal change. Therefore, the glucose level might be two times higher than normal without visible signs.
If someone wants to know their blood sugar levels, then a one-time measurement at the hospital also can be misleading.
Even today’s popular glycated hemoglobin test (A1c) shows only the average blood glucose level over the past three months. And if it constantly goes up and down, the test can give you an acceptable mean value.
Sure, you may also check the volume of pancreas insulin production by taking a C-peptide test. But that will not help you to monitor glucose levels daily.
Therefore, a smartwatch with a sugar-level measurement feature would be just a magical solution to many problems. And we will talk about “magic” a little later, but for now, we have to look at CGM systems that have become popular over the past decade.
Continuous blood sugar monitoring with CGM
CGM systems (Continuous Glucose Monitoring) can partially solve the problem of diabetics. These devices usually consist of a small disposable sensor and a radio transmitter.
A tiny sensor is attached to the body using an adhesive patch, and a wire electrode is painlessly inserted under the skin of the arm or abdomen. A transmitter is attached to the sensor and sends data wirelessly to the receiver (i.e., a smartwatch or a smartphone).
Here is how it looks:
This way, you can monitor your blood glucose levels in real-time. And what’s more, the system can automatically warn you when hypo- or hyperglycemia occurs.
In general, CGM systems measure sugar levels every 5 minutes with an error of about 10%5.
The apparent advantage of CGM systems is that you not only get a number taken out of context but also see the glucose levels changes over time and other meaningful data:
But unfortunately, CGM systems have their flaws.
The fly in the ointment
First of all, such devices measure glucose levels using an electrode under the skin or even an implanted sensor.
And when you insert anything into the body, it causes a rejection reaction. To prevent that, you need to use a biocompatible material. But even so, it will be coated with a layer of protein very soon inside your body, which will reduce sensor response and measurement accuracy.
So you have to either manually recalibrate the system or replace the sensor frequently. Almost all CGM systems will not allow you to use one sensor for more than 7-14 days (entirely implanted sensors can work up to 90 days without replacement).
Besides the obvious inconvenience, it results in additional expenses since one sensor costs about $70. And you need four of them every month!
But there are other drawbacks as well. CGM systems do not measure blood glucose; they take readings from interstitial fluid (a thin layer of fluid surrounding the tissue).
Rapid changes inside cells and blood vessels are not accompanied by similar changes in the interstitial fluid but follow with some delay. Such time delays (5-10 minutes20 on average) hamper the detection of hypoglycemic events. And values may vary since the correlation is indirect6.
Furthermore, when you move or change the position of your body, the sensor can come into firm contact with tissue which restricts or even completely cuts off the electrode access to interstitial fluid. As a result, you receive a false alarm for hypoglycemia.
And of course, the vast majority of healthy people will not bother with the purchase of expensive equipment and the frequent sensor replacement to monitor sugar levels.
We need a tool that makes it easy to check blood sugar whenever you want without puncturing the skin or implanting sensors into the body. We need a tool that does not require consumables.
You just have to press the button on your smartwatch and get the results. Only then everyone will want to check their glucose and do it periodically, which can significantly improve the situation worldwide.
A smartwatch with a glucose monitor. A pipe dream or a daring challenge to science?
So the challenge before the manufacturer is to make a device that could measure blood sugar without puncturing the skin or interacting with blood or even interstitial fluid. But how on earth is that possible?
Actually, we have several options here!
The first thing that comes to mind is to analyze other fluids, e.g. tears, sweat, saliva. And, indeed, such attempts have been made several times.
In 2014 Google developed contact lenses with a sugar measurement feature but cancelled the project after four years of stagnation. The official blog post7 states that it was an impossible task since the correlation between blood sugar and tears sugar is not strong enough for such lenses to be certified as a medical device.
Anyway, this method is not suitable for a smartwatch. So, we need entirely different technology, and such technologies exist, indeed, allowing (at least theoretically) to analyze glucose levels. But in reality, everything is far from being as simple as it might seem. So let’s talk about it.
If Apple or any other company ever releases a smartwatch with a blood sugar monitoring feature, it will be using one of the following technologies: reverse iontophoresis, molecular spectroscopy, or Raman spectroscopy. Let’s discuss each of these more in detail.
Iontophoresis is a technique of transferring ions (molecules with an electric charge) through the skin using an electric current.
We use iontophoresis to deliver a medicine or other chemical through the skin. But reverse iontophoresis allows us to do the opposite – remove molecules from within the body for detection.
Here we have a pretty straightforward method for measuring blood glucose. All we need to do is pull glucose molecules through the skin to the smartwatch sensor and count them!
We need to attach positive and negative electrodes to the skin and then apply a small current. The electric field will attract negatively charged molecules to the positive electrode and positively charged molecules to the negative electrode.
The only problem is that glucose has no charge, since it’s a polar molecule (polar molecules always have a positive charge on one side and a negative charge on the other). And therefore, it doesn’t care about the electric field created by such a smartwatch.
But this method does work because there are many other charged molecules in the interstitial fluid, such as sodium or chlorine (technically, these are not molecules but chemical elements). So these ions move due to electric field and create a flow of fluid with glucose and other non-charged molecules (see the electro-osmosis effect)8:
It’s the method used in the world’s first non-invasive glucose monitor to measure blood sugar without puncturing the skin. It was an infamous GlucoWatch, released 20 years ago and went bankrupt a short time later.
To extract a sufficient number of glucose molecules, GlucoWatch had to raise the voltage on the electrodes that caused skin burns. Also, it was necessary to change the disposable sensor worn beneath it every day and then recalibrate the sensor.
When GlucoWatch detected a sweat, it stopped working or gave a huge error. But the main problem was the accuracy of the measurements when the blood sugar levels dropped (hypoglycemia). The watch could not collect enough glucose even when there was a lot of it.
But it’s been a while now. The sensors’ quality and sensitivity have increased many times over the past 20 years; also, we’ve seen a neural networks boom. Therefore, we could see new non-invasive glucose monitors (non-invasive means without drawing blood, puncturing the skin, or causing pain) based on reverse iontophoresis technology.
The latest reverse iontophoresis glucose monitors show high accuracy (~12% error) when the blood glucose levels are >4.4 mmol/l. But when glucose drops, the error increases to 19% (>3.3 mmol/L) or 27% (>2.2 mmol/L)9 .
This is the broadest range of methods for molecular analysis. At the same time, the most convenient since the device does not have any consumables. All these methods are based on the analysis of molecule interaction with light (“light” means electromagnetic radiation in general, not just the visible spectrum).
Simply put, all these green, red and infrared LEDs on the back of smartwatches for measuring blood oxygen are a kind of spectroscopy:
I think most users believe that the glucose monitor on a smartwatch will work just like a pulse oximeter (for measuring blood oxygen). Maybe so, but there are many challenges to solve.
Smartwatches with glucose monitors will use one of these technologies:
- NIR spectroscopy (in the near-infrared spectrum)
- Raman spectroscopy
Speaking of infrared NIR spectroscopy, the blood glucose measuring is as follows: the LED shines onto skin with infrared light that has a frequency ten thousand times higher than used in 5G technology.
When this invisible light hits a molecule, it can partially be absorbed, scattered or reflected. The energy of the IR spectrum is too small to be absorbed by molecules’ electrons. So the glucose molecule absorbs IR light by its rotation, vibration or stretching:
Since all molecules have a different structure, each absorbs and reflects its part of the spectrum. The incident light must be the same frequency as molecule vibrational frequency.
For each material, we will have a different light spectrum. If we know the “spectrum” of glucose molecule, we can measure the amount of glucose by measuring the reflected light intensity.
That is the essence of IR spectroscopy. Near-Infrared light easily penetrates deep into the skin and interacts with tissues. Part of the light reflects, and the neural network analyzes its spectrum and intensity. Knowing the spectrum, we understand that it’s glucose and then calculate its amount by intensity.
Sounds pretty simple, right?
As I said before, this method is similar to the one used when measuring the blood oxygen level, but there is a vast difference.
First of all, there are 140 times10 more hemoglobin molecules (for oxygen transport) than the amount of glucose inside the human body. Accordingly, it’s much easier to work with such an amount of a substance than with glucose.
Also, hemoglobin, combined with oxygen and hemoglobin without oxygen, have different colors. And it’s easy to distinguish their spectra since they absorb electromagnetic waves at different frequencies. But the glucose molecule is entirely colorless.
Besides, many molecules have structures similar to glucose. Therefore their signals will overlap, making it very difficult to distinguish glucose from other compounds produced by glucose metabolism.
But there’s more! When we emit infrared light on the skin, a significant part of it will immediately reflect from the skin’s surface and return to the receiver (photodiode). This is a strong but completely useless signal that does not contain any information about glucose. It’s actually noise.
Even if you filter out the noise and identify a weak signal associated with glucose, how will you distinguish a drop in glucose levels due to diabetes from the constriction of blood vessels due to a temperature drop?
After all, when it’s cold outside, the body tries to narrow the vessels in the limbs, thereby reducing the blood volume. And since the vast majority of glucose is in the blood, the sensor will notice this drop. But it has nothing to do with the disease.
All these problems need to be solved somehow before we see a smartwatch with a built-in high accuracy glucose monitor.
Furthermore, even a slight change in the strap tension will significantly distort the result since the excess pressure will affect the spectrum.
Alternatively, you can use the so-called Raman spectroscopy (named after the Indian physicist Raman).
Unlike IR spectroscopy with LEDs, this method requires a laser (light with a single wavelength). By comparison, light from a heart rate sensor consists of many different wavelengths, which we see as a single color:
To get the basic idea of how this method work, you need to remember that light can not only be reflected or absorbed but also scattered. Scattering is a phenomenon in which light collides with some particles and changes its properties (mainly, the direction of movement).
There are two types of scattering: elastic and inelastic.
In the case of elastic scattering, the light doesn’t change its frequency (energy), only direction. We can observe this type of scattering everywhere. For example, the sky is blue due to elastic scattering (tiny particles in the atmosphere scatter blue light in all directions, making the sky looks blue).
But with inelastic scattering, the frequency of light changes depending on a substance the light interacts with.
Now, let’s see how such a glucose monitor should work. First, the light from a smartwatch laser penetrates the skin. It collides with various substances, including glucose, and then scatters.
Since it’s inelastic scattering, we get a set of light waves with different frequencies. Smartwatch filters out the most powerful light scattered elastically without changing the frequency (useless light reflected from the skin) and amplifies the rest of the signal.
Since the Raman spectrum of glucose is well known (each substance has its spectrum or its own set of waves at different frequencies, like a “fingerprint”), it’s much easier to extract this information from the signal:
The amount of glucose directly depends on this spectrum’s intensity (“brightness”).
And what’s interesting, in 2013 Apple hired key employees of C8 MediSensors company11, which was developing a Raman spectroscopy glucose monitor that gave an error of just 2 mmol/l.
The major approaches of non-invasive glucose detection in a smartwatch are reverse iontophoresis, IR spectroscopy and Raman spectroscopy.
But there are many other possible techniques as well, including the photoacoustic effect (a laser beam irradiated on the skin produces a thermal expansion, thereby generating a sound which is affected by the molecule type), bioelectrical impedance analysis (a method widely used in smart scales for estimating body composition, in particular, body fat and muscle mass), etc.
So let’s summarize key points of this article:
- Everyone needs to know their blood sugar levels, if only to make sure they don’t suffer from diabetes mellitus (primarily type 2, which can be hard to identify).
- Measuring glucose with a smartwatch is not that crazy of an idea. In theory, there are many ways to do this without skin puncturing.
- Currently, there is no commercially available certified non-invasive glucometer that can be worn on the wrist.
I don’t know whether we’ll see a smartwatch with a glucose monitor this year. Still, I can assure you that such devices will not replace the “classic” glucose monitors or certified CGM systems for at least ten years.
There are a lot of challenges that need to be solved if we want to have a reliable tool and not just a gimmick device. And so far, no one has been able to do this, despite the millions of dollars that investors are willing to pay anyone who wants to give it a try.
Modern smartwatches may not always measure blood oxygen levels or even heart rate with sufficient accuracy, let alone blood glucose!
Alex Salo, Tech Longreads founder
If you enjoyed this post, please share it! Also, consider to support Tech Longreads on Patreon and get bonus materials!
- IDF Diabetes Atlas [source]
- Der Glucoseverbrauch des Gehirns und seine Abhängigkeit von der Leber [DOI]
- Avogadro constant [source]
- GOST R ISO 15197-20154
- Accuracy, Utilization, and Effectiveness Comparisons of Different Continuous Glucose Monitoring Systems [PMID: 30681379]
- Interstitial Fluid Glucose Is Not Just a Shifted-in-Time but a Distorted Mirror of Blood Glucose [DOI]
- Update on our Smart Lens program with Alcon [source]
- Electrochemical Sensors for Clinic Analysis [DOI]
- A Prospective Single Centre Evaluation of the Accuracy and safety of the sugarBEAT CGM System [source]
- The Pursuit of Noninvasive Glucose: Hunting the Deceitful Turkey [Book]
- iWatch’s novelty emerges as Apple taps sensor and fitness experts [source]
- Everything You Wanted to Know About Noninvasive Glucose Measurement and Control [source]