Non-Invasive Devices to Catch Overnight Lows (Pipe Dreams?)

By Mike Hoskins and Amanda Cedrone

We all know the fear that low blood sugars can bring, especially when it comes to the thought of going to sleep and not waking up because of hypoglycemia.

Many of us have found some comfort in using a CGM, alleviating lows, but there are still some problems with the affordability and accuracy of these devices. Two new companies are aiming to create solutions here.

Their products aren’t exactly CGMs, but rather are designed to detect low blood sugars only at night and in a way we’d all want to see in our dreams: non-invasively, without poking through our skin.

Of course, we all tend to roll our eyes whenever we hear about a “non-invasive” device — a pipe dream for many in the Diabetes Community. And it’s especially disappointing to learn that the first of these non-invasive night-time hypo monitors, developed overseas, has already been recalled in the first months it was on the market (!)

HypoMon

Known as the HypoMon that’s short for Hypoglycaemia Monitor, this D-device created in Australia dates back to about 1992. It was demo’d for the first time in the U.S. back in 2006 at the ADA Scientific Sessions and eventually hit the market in Australia and the UK last fall.

HypoMon is a wrap-around belt with a sensor and transmitter that measures electrical activity of the heart and movement information from the user in order to determine low blood sugars at night, according to the website. The data is transferred wirelessly to a monitor that must be within a 10-meter radius of the sensor, and if low blood sugar is detected, the monitor will sound an alert.

It is not a CGM and does not provide the user with blood glucose readings, only alerts when the BG level drops beyond a dangerous threshold. Because the HypoMon is not intended for use other than nighttime, each monitoring session will automatically end 10 hours after it began.

This device is the creation of Professor Hung Nguyen of University of Technology in Sydney, Australia, who started using his background in bioengineering and robotics to explore non-invasive glucose monitoring through sweat, heart rate and ECG characteristics. He founded his company AIMEDICS in 2001 and brought that research on board, and they’ve been pursuing it ever since.

The company website states that the product is intended for type 1 PWDs between ages of 10 and 25, but it’s not exactly clear why it couldn’t be worn by anyone older or younger. The product was retailing for about $3,000, according to pricing information listed online, which seems somewhat steep to us, since CGM starter kits are about half the price or less.

After becoming available in Australia and the United Kingdom in November 2012, the product was voluntarily recalled by AIMEDICS at the beginning of this month, “because it was not performing as well as expected.” We have no idea what that means, and HypoMon did not return requests for an interview. But the AIMEDICS chairman is quoted on the HypoMon website saying, “The AIMEDICS board has made this voluntary recall decision as we want to investigate further technical options to improve the HypoMon user experience.”

According to the Australian Therapeutic Goods Administration (Australia’s version of the FDA) HypoMon’s rate of detection of sleep-time hypoglycemia episodes is lower than the rate specified in HypoMon’s instructions for use.

Those instructions say: “The expected performance in a 30-day month, where the user experiences between 10 to 15 nocturnal hypoglycaemic episodes, or ‘hypos’, on different nights, the HypoMon is expected to alarm on 8 to 12 ‘hypos’… It is expected that in the same 30-day month, the user may typically experience between 15 to 20 nights without a ‘hypo’. It may be expected that the HypoMon produces an incorrect or ‘false’ alarm on 3 to 4 nights. False alarms occur on 20% of nights without a ‘hypo’, giving a device specificity of 80%.”

Disappointing — to say the least! But the company hasn’t folded, and they seem to intend to keep up the fight for a viable product.

DropSense

A second product underway for the same purpose as HypoMon is called DropSense — currently being developed by a small group of University of California at Berkeley students. It’s a non-invasive device that will use a biosensor and algorithms to determine and subsequently alert users to overnight low blood sugars via a mobile application. Like HypoMon, it doesn’t give the user blood sugar readings; it just alerts the user when they have symptoms indicating that blood sugar may be dropping. And also like HypoMon, it is only intended for nighttime use.

DropSense is the brainchild of 20-year-old PWD Steve Yadlowsky, an electrical engineering and computer science student at UC Berkeley who has been living with type 1 for almost 13 years. He’s been using a Dexcom CGM for five years, and his medical insurance coverage comes through his parents’ insurance. Last year, there was talk of the employer switching insurance companies, and Steve feared that this new insurer wouldn’t cover the costs of monthly CGM sensors and he’d be left footing the bill.

“If you look at how much sensors cost, it’s equivalent to the monthly cost of leasing a BMW,” he said.

The idea of DropSense came from that insurance coverage concern; basically, it led Steve and his fellow students at UC Berkeley to start thinking of lower-cost options to alert for low blood sugars.

Still in early stages, DropSense will be a non-invasive sensor that either sticks to your body or you wear like a wristband, that connects to a mobile app communicating via smartphone or iPad. Unlike a traditional CGM that gets inserted under the skin and reads a PWD’s interstitial fluid, this tech would detect individual physiological responses like increased heart rate and sweating — two of the more common signs of a hypo. The sensor could identify and detect those specific physiological responses to low blood sugars, and then an algorithm being developed would interpret the data in order to send alerts to a smartphone when those symptoms are detected. Eventually, the creators would like that to be a predictive technology.

Considering the HypoMon recall, and our inherent skepticism of non-invasive products, we had a lot of questions for Steve about how his product is different than other attempts at non-invasive monitors that haven’t panned out at all. What differentiates their effort?

“One thing is simply timing. Mobile technology puts an incredibly powerful and sophisticated computer in billions of people’s hands. With the recent breakthroughs in wearable computing there have been vast improvements in cost and accuracy of biosensing technology. We leverage both of these to create an advanced system for low blood sugar detection. Without either of these pieces, our approach would be impossible, which explains why no one has done it this way before,” Steve said.

He says these advances in technology allow DropSense to be data-driven, using learning algorithms that can help predict behavior and recognize it better in the future.

As Steve describes it, this technology would like the low-blood sugar version of the iPhone “Siri.” The algorithm watches signals that the group reads from a sensor and uses them to make a prediction about the chance that the user is experiencing, or about to experience, a low blood sugar based on patterns.

“Similarly to how Siri can recognize whether you said ‘call’ or ‘text’ based on sound data, our algorithm can distinguish lows apart from rolling around in one’s sleep using our sensor data,” Steve said.

Steve is focusing on detecting low blood sugars instead of high because research shows that fear of lows significantly impacts overall diabetes care. The first version of the product will most likely be intended solely to detect nighttime lows and will not be tested for use throughout the entire day, Steve said. However, all-day use is a concept the team plans to look into in the future.

He’s also had a few scary overnight low experiences of his own. Prior to leaving for college, Steve had a few instances where his blood sugar dropped to the 30s while sleeping. He didn’t realize until his parents heard his CGM alarm and woke him.

“Close calls like that that made me realize this is solvable,” he said.

Steve said the team has developed a few prototype sensors, and he’s been testing those out on himself. They’re planning to do a larger trial soon, with about 50 patients, that will allow the team to accumulate data, understand how well the algorithms are working and make adjustments accordingly. The group has not made contact with the FDA and doesn’t have any sort of product development timeline yet.

That being said, getting the product to market as quickly as possible is a priority, Steve said. And while FDA clearance on some products can take years, he points out that the FDA has recently been working with mobile app and wearable technology developers to improve accelerate the process of evaluating innovative medical devices, and that clearance takes less time for companies that are well-prepared for audits conducted during the clearance process.

The group doesn’t have an estimated price for the product yet, saying only that it would be more affordable than a CGM. DropSense is currently self-funded and looking for investors (of course!)

Expert Skepticism

Will a device like this ever work well enough to really be reliable? It’s tough to imagine, given the quick recall of HypoMon’s first attempt.

It’s also tough to tell how popular a night-time only device like this would be if it were widely available. It would have to be significantly cheaper and easier to get than a CGM, since the benefit is just an “emergency alarm system” for overnight.

We asked diabetes technology expert Dr. Barry Ginsberg, a veteran source on glucose monitoring tech, about his reaction to the HypoMon recall and this type of non-invasive low-detecting device. He wasn’t surprised the HypoMon product didn’t work as expected, and he’s skeptical about the need and feasibility overall.

The biggest problems he sees with these type of products are that they give a lot of “false positives,” and PWDs who arehypoglycemia unaware (about 20% of type 1s) don’t always give off the physiological symptoms that the products are testing for to indicate low blood sugar. Basically, the bodily signs that these products are looking for to indicate low blood sugar don’t exist. So what’s the use?

On DropSense, Ginsberg said that while he isn’t familiar with the technology that Steven and his team are using, he’s skeptical for the same reasons. In fact, he questions the need for a night-time low blood sugar alert system product altogether.

“I don’t see the future for us in things like HypoMon. The direction I see is continuous monitors,” he said, noting that lower-cost CGMs are worth exploring. “The current CGMs, the best of them, the inaccuracies are 5 or 6 percent. Do you know how much it costs if you’re low and have to go to the emergency room?!”

Right. The technology is fascinating, and we’re excited to see where anything “non-invasive” goes, but we’re not putting our money on these gadgets either. Instead, we’ll keep using the best-of CGMs we’ve got now, and keep our hopes up that non-invasive dreams eventually do come true.