Managing Diabetes: Why Smartwatches Are Not a Replacement for Insulin Needles

Bloomberg recently published an article that created a buzz in the health tech industry by revealing that Apple had achieved a significant breakthrough in noninvasive blood glucose monitoring. This breakthrough has the potential to revolutionize the treatment of diabetes. However, despite the hype, it is unlikely that this technology will be available on the Apple Watch or any other consumer wearable for several years. There are technical and regulatory obstacles to overcome for emerging health tech, and even if they are resolved, experts believe that noninvasive blood glucose monitoring may not replace traditional finger prick tests. Furthermore, the most realistic and helpful use for this technology may not even be for replacing traditional tests.

Pinprick-Free Testing

Noninvasive blood glucose monitoring refers to the measurement of blood sugar levels without the need for invasive procedures such as drawing blood or causing pain or trauma to the skin. Pursuing this technology is worthwhile for several reasons, with the main one being its potential to treat diabetes. Diabetes results from the body’s inability to regulate blood sugar levels effectively, caused by a lack of insulin production (Type 1) or insulin resistance over time (Type 2).

Currently, both Type 1 and Type 2 patients manage their condition by checking their blood sugar levels through invasive methods like finger prick tests or continuous glucose monitoring (CGM). Finger prick tests involve pricking the finger with a needle and placing a drop of blood on a test strip, while CGM involves embedding a sensor underneath the skin to monitor blood sugar levels in real-time 24/7.

Few people enjoy getting poked with needles for yearly shots, let alone daily glucose checks. So you can understand the appeal of noninvasive monitoring. Patients wouldn’t need to draw blood or attach a sensor to their bodies to know when they should take insulin or monitor the efficiency of other medications. Doctors would be able to remotely monitor patients, and that, in turn, could expand accessibility for patients living in rural areas. Beyond diabetes, the tech could also benefit endurance athletes who have to monitor their carbohydrate intake during long races.

It’s one of those scenarios where everybody wins. The only problem is that research into noninvasive blood glucose monitoring began in 1975, and in 48 years, nobody’s been able to figure out how to reliably do it yet.

Locating the Glucose Signal Amidst the Biological Haystack

ight now, there are two main methods of measuring glucose levels noninvasively. The first is measuring glucose from bodily fluids like urine or tears. This is the approach Google took when it tried developing smart contact lenses that could read blood sugar levels before ultimately putting the project on the back burner in 2018. The second method involves spectroscopy. It’s essentially shining light into the body using optical sensors and measuring how the light reflects back to measure a particular metric.

If it sounds familiar, that’s because this tech is already in smartwatches, fitness trackers, and smart rings. It’s how they measure heart rate, blood oxygen levels, and a host of other metrics. The difference is, instead of green or red LEDs, noninvasive blood glucose monitoring would use infrared or near-infrared light. That light would be targeted at interstitial fluid — a substance in the spaces between cells that carries nutrients and waste — or some other vascular tissue. As with heart rate and blood oxygen, the smartwatch would theoretically use a proprietary algorithm to determine your glucose levels based on how much light is reflected back.

But while the method is similar, applying this tech to blood glucose is much more complicated.

“The signal that you get back from glucose happens to be very small, which is unfortunate,” says David Klonoff, medical director at the Diabetes Research Institute at Mills-Peninsula Medical Center in San Mateo, California. Klonoff also serves as president of the Diabetes Technology Society, editor-in-chief of the Journal of Diabetes Science and Technology, and has followed noninvasive glucose monitoring tech for the past 25 years.

When it comes to glucose, it turns out size matters. That small signal makes it difficult to isolate glucose from other similarly structured chemicals in the body. It’s a headache for device makers, who can get tripped up by something as simple and ubiquitous as water.

“Water interferes with measurement in optical methods, and our bodies are filled with water. If you have any subtle changes in amounts of water, that can dramatically affect the signals you’re measuring,” says Movano CEO John Mastrototaro. Movano made waves for developing a women-first smart ring at CES, but the company has also developed a chip that may potentially be able to measure blood pressure and blood glucose using radio frequencies.

Both Klonoff and Mastrototaro also noted that substances within the body aren’t the only things that make isolating the glucose signal difficult. External and environmental factors like stray light, movement, and poor skin contact with the sensor can also throw off noninvasive measurements. Plus, infrared light is essentially a form of heat. It’s invisible to the naked eye, but all objects — including humans — give off some kind of infrared heat. And sensors aren’t always able to tell whether that heat’s coming from your smartwatch or a sweltering summer day.

Poor skin contact, movement, and stray light can throw off measurements. Image: Vjeran Pavic / The Verge

For example, say you’re living in a future where smartwatches can noninvasively monitor your blood sugar levels. Climate change triggers a massive heatwave, and your HVAC breaks down. The room gets hotter, you get sweaty, and your smartwatch’s sensor could easily mistake that extra heat as your blood sugar rising.

One workaround is to collect more data by using multiple wavelengths of light — as in, adding more sensors that emit different types of infrared light. The more you have, the easier it is to figure out what’s glucose and what’s interference. But stuffing in more sensors comes with its own set of issues. You need a more powerful algorithm to crunch the extra numbers. And if you add too many wavelengths, you risk adding more bulk to a device.

There are sensors small and power efficient enough to fit into a smartwatch, but taking frequent, continuous measurements will still drain the battery. For example, many wearables that support nighttime SpO2 tracking will warn you that it may dramatically lessen battery life once the feature is enabled.

Current CGMs take measurements roughly once every five minutes, so a noninvasive smartwatch monitor would need to at least match that while maintaining at least a full day’s worth of battery. It has to do that plus track activities, power an always-on display, measure a host of other health metrics, fetch texts and notifications, and send data over cellular or Wi-Fi — all this without resorting to adding a bigger battery so the device can be comfortable enough to wear to sleep for truly continuous monitoring.

The accuracy of optical sensors may vary for individuals with darker skin tones or tattoos.

Another potential issue: optical sensors may not be as accurate for people with darker skin and tattoos. That’s because darker colors don’t reflect light in the same way as lighter colors. Take pulse oximeters, which use red and infrared light to measure blood oxygen. An FDA panel recently called for greater regulation of these devices because they were less accurate for people with darker skin. Noninvasive blood glucose monitors may not have as big of a problem here, as infrared light is better at handling melanin and ink than visible light. But even with that advantage, Mastrototaro says it’s still a challenge with wavelengths currently used in noninvasive glucose monitoring.

Regulatory clearance means adjusting expectations

Despite all of these challenges, technology has evolved to the point where many of these are solvable issues. AI is more powerful, so building algorithms that can handle the complexities of noninvasive glucose monitoring is easier than it used to be. Chips and other components keep getting smaller and more powerful. Companies like Movano are actively exploring alternatives to optical sensors. But technology is only one part of the equation.

There’s also the FDA.

Wellness features, like blood oxygen spot checks or heart rate, don’t require the FDA to weigh in on safety or efficacy because they’re for your own awareness. But the stakes for blood glucose levels are much higher. An incorrect reading or false alarm could lead a Type 1 diabetic to administer the wrong dosage of insulin, which could result in life-threatening consequences. For that reason, any smartwatch touting blood glucose monitoring features would have to go through the FDA.

The rub is obtaining FDA clearance or approval is a laborious process that takes months if you’re lucky and years if you aren’t. Device makers have to conduct rigorous testing and clinical trials for accuracy, safety, and efficacy. As frustrating as this is for companies, this level of rigor is a good thing and protects us, the consumers. But there’s no guarantee that any company — even one with a really good idea — will successfully make it through the process. And for many, that’s not a bet worth taking if the pros don’t significantly outweigh the cons.

This is why it’s extremely unlikely that consumer tech companies will even try to replace established methods like the finger prick test or CGMs, at least not anytime soon. It’s more likely that blood glucose on smartwatches will be for fitness or wellness tracking or, more ambitiously, a screening tool for prediabetes.

It’s more likely that blood glucose on smartwatches will be for fitness or wellness tracking

It’s essentially the path every wearable maker has followed thus far. When Apple introduced FDA-cleared EKGs on the Apple Watch Series 4, the purpose was to flag irregular heart rate rhythms and suggest you see a doctor to assess your risk of atrial fibrillation. It was never intended to help you manage a condition or inform treatment. Other companies like Fitbit, Samsung, and Garmin do the same for their EKG and AFib detection features.

These kinds of screening features may not sound quite as revolutionary, but they create a win-win scenario for researchers, companies, and consumers alike. In this case, the CDC says 96 million American adults have prediabetes, while Type 2 makes up 90 to 95 percent of diagnosed diabetes cases. It’s cynical, but this population represents a bigger customer base for companies for a lot less risk. Plus, all the data gathered from noninvasive monitoring could lead to new insights for researchers and consumers.

“I think what we’re going to see is that there’ll be subtle patterns that we don’t recognize right now that will alert people that they’re somewhere between normal and diabetes. And I think there are going to be patterns that predict certain types of prediabetes,” says Klonoff.

“It’s not just knowing your glucose that’s important. It’s really understanding everything about your health,” adds Mastrototaro, noting that, if successful with its RF tech, Movano hopes to fold glucose into its platform alongside other health metrics like heart rate, activity, and blood oxygen. That, he says, is more valuable as it creates a more complete picture of a person’s health. It’s also the same approach that Mastrototaro took back at Medtronic, where he worked on the team that made the first FDA-cleared CGM in 1999.

“Basically, the tool of the CGM allowed you to monitor trends in people’s glucose over time, so kind of to get an idea of the big picture. That’s where we started and we weren’t using it for real-time monitoring,” Mastrototaro explains, referring to how a Type 1 diabetic may use CGMs to determine how much insulin to take. “In the labeling of the initial products, it said that you can use this data for trends, you can use it to give you an idea, you can even use it to alert you if it thinks your blood sugar’s going too high or too low, but then you should confirm it with one of the fingerprick tests to verify and then treat.”

Sounds an awful lot like how smartwatches detect irregular heart rate rhythms before advising users to seek an official diagnosis from a doctor.