DiabNext Brings Artificial Intelligence to Diabetes

What if we could have an ultra-cool Artificial Intelligence platform like the one Marvel comics invented for Iron Man that would combat our diabetes the way it did villains?!

Well, you guessed it: a med-tech company in the real world is bringing their own version of Iron Man‘s J.A.R.V.I.S. interface — or Just A Rather Very Intelligent System — to the diabetes space.

Meet DiabNext, whose name as you might imagine is a mashup of Diabetes and Next- generation. This company has developed a smart connected insulin dose tracker called Clipsulin that’s compatible with all insulin pens, captures the data and funnels it into its still-in-development AI platform called JARVIS on either a smartphone or computer.

So intriguing! Could it be nearly as cool as what the fictional Tony Stark created to control his Iron Man mansion? More on that below, but first a bit about DiabNext…

Remember the Bee smart insulin pen cap that tracked insulin dosing data, first introduced in 2014? Well, this is the same product then offered by Switzerland- based watch company Vigilant and was eventually referred to as the VigiPen. But as of 2017, the company name’s been changed to DiabNext and the insulin pen recorder itself is called Clipsulin. This same company is known as Shenguo Technology in Hong Kong, which incorporates Chinese into the name. Carrying over from Vigilant as general manager is the same T1 PWD, Laurent Nicolaus, who was diagnosed in his 30s more than a decade ago.

But company names and branding aside, the interesting thing here is the technology itself. To get the scoop on that, we talked recently with Sam Chen, DiabNext’s director for the USA and Canada, based in Boston.


Their next-gen smart insulin pen tracker is known as Clipsulin — because it clips to your insulin pen, of course. That is, the main design change from earlier models is that this it’s no longer a cap replacement, but a smaller tool that clips to the side of any insulin pen.

“Bee is gone and we’ve gone through a big transformation,” Chen says. “There wasn’t an option to collect data, so we created that with Clipsulin. It’s not a super-complicated item, but is really doing for people what they’d otherwise have to manually do by writing (results) down on paper or notepad. This makes it easier.”

No more manual-entry of your insulin doses either, as Clipsulin will record that automatically once you dial in your dose. There is an LED screen that detects the dose and displays it, and then via Bluetooth (or other infrared signal) sends that data to an Android or iOS app.

OK, let’s keep it real here, folks: this isn’t all too different from what’s already available or on the horizon in smart insulin pen tech — from the NovoPen Echo (working only with that brand of insulin pen), to the Timesulin tracker that’s going Bluetooth, and others coming soon like the FDA-cleared Companion Medical InPen and the Emperra ESYSTA pen.

Chen claims that Clipsulin doesn’t need the FDA’s OK since it’s not going to be used for actual dosing decisions, but rather just tracking and displaying the data within a mobile app and bigger cognitive platform eventually. This seems odd to us given that the original Bee insulin pen tracker and others have been subject to FDA clearance…?

DiabNext is apparently planning for a global launch of this universal clip within 2017. Chen also says that last year, the company signed an agreement with Sanofi to create a specific version of Clipsulin for their insulin pens available in Germany and elsewhere overseas.

What will the cost be? Well, DiabNext doesn’t know yet. Since it’s all connected to the bigger AI piece that’s still in the works, we’re told.

“This is just one piece of our larger platform puzzle, which will connect into our AI,” Chen said.


What exactly sets Artificial Intelligence (AI) apart from all the other computer programming and algorithms being built into products today? It’s a form of intelligent programming that allows the system to learn and problem-solve, emulating human thinking. Experts agree that AI is the wave of the future in medical practice, and diabetes care in particular. In fact, last year’s European Conference on Artificial Intelligence (ECAI) featured a whole workshop on Artificial Intelligence for Diabetes with dozens of researchers discussing portable personalized decision support systems for insulin dosing that combines data from multiple sources such as body-worn sensors and manual inputs. (DiabNext wants to do away with the manual part.)

One leading endocrinologist from Nassau University predicts that “within 20 years most of diabetes will be managed by AI algorithms and smart machines.” He describes the fit between diabetes care and AI quite nicely:

“The core of diabetes management is pattern recognition. At present, physicians look at blood glucose logs and spots these trends and make treatment recommendations. The key factors that affect these recommendations are desired level of diabetes control, weight of the patient, food habits, other medical problems, current medications and affordability. But we know that genetic variations can influence how we respond to drugs. At present, it is humanly impossible to memorize all those genetic variations and prescribe medications. If we feed AI enough data, it will be able to make better judgement calls on diabetes management.“


DiabNext’s JARVIS interface is designed to allow doctors and patients alike to “tap into the power of AI and supercomputing from any internet connected tool.”

The officially description states: “In particular, physicians logged in… can for the first time visualize the patients’ insulin injection therapy metrics, oral medicine intake, blood glucose levels, meals and precise carbs intake calculation, diabetes standard diagnostics test results, A1C trends, diabetes- related genes sequencing profile, and even workout and weight data trends to better assess what drives patients’ highs and lows. Physicians, Nutritionists, Caregivers, Researchers and Patients are ONE TEAM in DIABNEXT® A.I., designing together the therapies of the future to prevent and cure Diabetes for our generation and our children.”

Sound like a lot of hype? Chen explains: “Our AI will not be diagnosing or curing you, but will be more like a personal assistant you can talk to for your data. It can smartly interact with you, and the more you use it the more you get from it. It learns from human interaction. Like a partner in your diabetes.”

He clarifies that JARVIS will not only collect the insulin dosing data from the Clipsulin tracker, but will bring together glucose meter data from Bluetooth- meters being used as well as activity and carbohydrate data trackers. The system will allow users to snap photos of food to get image recognition for carb counts. (OK, that’d be cool!)

“If you’re in a hurry and just want to ballpark it, you can do so quickly based on the database carb count instead of changing it to be more exact. That’s our vision, of allowing things to be done quickly so you don’t have to do it all manually.”

Chen notes that this ‘personal assistant’ approach is quite different from other news-making D-technology, such as the IBM Watson-powered Medtronic-introduced SugarIQ app that “provides context and insights” based on insulin pump and CGM data.

For example, with the DiabNext system, if the user inputs a need to increase their activity level, the AI will respond with asking about the type and level of activity, and based on that, will use GPS to locate nearby gyms or places where you can go for that exercise. The same would apply to food, and over time it can learn about your patterns — such as types of pizza you eat and the subsequent BG effect based on carbs, insulin, and glucose readings — to better help you manage diabetes.

“It can continue to learn, training with pictures and interaction that gets smarter and smarter,” Chen says.

While DiabNext has explored the idea of a coaching service within this platform, that’s not something it has in store at the moment. They are working with dietitians and could weave nutrition advice aspects into JARVIS, but would then refer PWDs to their actual medical care team for more in-depth discussion and follow-up. DiabNext also isn’t ruling out partnerships with other D-tech companies, gyms, clinics, or CDEs.

“The whole idea is to make life easier with diabetes, to integrate with doctors and make this a one-stop shop for anything you need in diabetes management,” Chen says. “We’re thinking this AI can help those with prediabetes, either to find out and even if they’re trying to reverse (the symptoms).”

As to cost, Chen says DiabNext is talking with payers to determine the level of coverage, and they’re also exploring the subscription-based model, but haven’t made any final decisions.

The whole concept is interesting, and if it does anything close to what they describe, it will certainly be innovative.

We’ll just have to wait and see how quickly this superhero diabetes assistant materializes…

This content is created for Diabetes Mine, a leading consumer health blog focused on the diabetes community that joined ishonest Media in 2015. The Diabetes Mine team is made up of informed patient advocates who are also trained journalists. We focus on providing content that informs and inspires people affected by diabetes.

Read more on: diabetesmine