Believe it or not, we've reached a point where a smartwatch worn on your wrist can actually recognize whether you're eating or drinking, and even track quantities and the speed at which you're consuming!
Just imagine the possibilities for helping manage diabetes... the device could instantly send you an insulin dosing reminder right in the very moment you open your mouth for the first bite.
This functionality is happening for real, introduced this week by a digital health startup named Klue in San Mateo, CA, that was founded in 2016.
Their technology tracks hand-to-mouth movements using the already-existing sensors built into smartwatches and off-the-shelf wearables like the Apple Watch via a mobile app. These hand gestures provide clues (hence the name 'Klue') to when someone is eating or drinking -- though not necessarily what kind of food or beverage.
As just announced at our DiabetesMine University innovation event late last week, Klue is also offering a real-time insulin bolus-reminder function to help PWDs (people with diabetes) manage insulin doses based on when they’re actually eating.
“There’s an enormous amount of valuable information encoded in our hand gestures about our behaviors, and today all those insights are basically untapped. So we developed an analytics technology that allows us to automatically -- without any user intervention, detect and track behaviors -- starting with eating and drinking," explains Klue CEO and co-founder Katelijn Vleugels, who’s lived with type 1 herself for many years.
"Don't we all want our lives back, where don't have to think about doing anything and the technology can regulate it for us? That's the Holy Grail, at least for me. I think with Klue, I'm a strong believer we can get there," she adds.
Following the Klue demo at the DiabetesMine event, Dr. Bruce Buckingham of Stanford Diabetes Research Center got up and made impassioned remarks about how this technology could aid future Artificial Pancreas systems. "Meal announcements are kind of the missing link," he said. "Klue has huge potential to automate that so the system would know when and how much to bolus."
Tracking Hand-to-Mouth Movements
You may not be aware that movement-detecting sensors are already built into wearables, and Klue simply taps into those sensors to analyze and feed the data into its mobile app on a person’s smartphone.
Based on Artificial Intelligence (AI), the Klue tech can decipher subtle differences in hand gestures, recognizing whether you’re eating or drinking and tracking data such as when, how often, and how fast the hands are moving to the mouth. That data fed into the smartphone app is synthesized into a Personalized Consumption Graph, showing users when and how fast they’re eating – a way to identify your own eating patterns and behaviors that might influence your blood sugars and food and dosing choices.
The system also offers real-time coaching, on eating speed and staying hydrated -- in the form of "nudges using little moments, many times."
This unique ability to detect behaviors in real-time, which engages people right in the moment of choice and action, has led the company to bill Klue as "the world's first operating system for behavior change."
Sharing Food Consumption Data
Since it works on existing platforms such as Apple Health, the Klue data can be easily shared and integrated with other health data in the same platform.
While Klue does not yet have any capability to weave in data from a CGM or glucose meter, or to add in manual entries or even photos of food choices, these functions are certainly possible for the future.
Klue’s Vleugels says they are eager to work with partners, including diabetes device makers and other digital health and app companies, that could help expand on future capabilities of this hand-gesture technology.
Beyond just diabetes, Klue is focused on overall health behavior change when it comes to food. The startup has a mobile app that's currently available through employers and payers, and employees using the Klue app get a license code to start using it and tracking data.
In Spring 2018, Klue unveiled partnerships with Stanford University and Crossover Health, a provider of employer-sponsored healthcare. The collaboration with Stanford is led by Christopher Gardner, a professor of medicine and director of nutrition studies at the Stanford Prevention Research Center. The entities worked together on a five-week clinical study to look at Klue’s application. The startup also teamed up with Crossover Health in the company’s Bay Area clinics to study the effects of this wearable tracking technology through employee use.
Diabetes Reminder Features
The new bolus reminder functionality is an exciting addition specifically for insulin-using PWDs.
"If you're living with type 1, there's no greater time you are thinking about this than when you start eating," Vleugels says. "That's when you have to make some of the big decisions, and there's so much you are thinking about at that time. It turns out that 25% of T1 adults miss or are late in bolusing for meals, and missing insulin doses is one of the main contributors to poor glycemic control."
At the moment it detects hand gestures indicating eating, Klue's new reminder module on Apple Watch offers a pop-up a message asking the PWD if they are indeed eating and want to bolus insulin. Obviously, it's just an alert. You would still need to use your pen, pump or syringe to dose the actual insulin. The Klue app also allows you to "snooze" the bolus reminder for 60 minutes if you aren't interested in bolusing then but don't want to forget completely. If you are eating specifically to treat a low blood sugar, you can tag the "hypo" option in the pull-down menu so that it is captured in your data tracking as a low event.
It also has an option for users to send automatic text message notifications to parents or caregivers each time a meal is detected.
These are the current features, but Vleugels points out that many more will come, as they work with partners in creating additional modules. At our DiabetesMine #DData18 event, Klue opened up a private beta testing to the general public to sign up and get limited (first come, first serve) access to this Klue app and technology: goklue.com/T1d.
A Personal Diabetes Connection
For Vleugels, a former Stanford associate professor who hails from Belgium, this is personal, given her own type 1 diabetes diagnosis as a young adult. She tells us she has always considered herself “fortunate” as an engineer and someone who loves math, and that prepared her for the challenge of T1D. While she's successfully been able to run those "complex D-Math algorithms" in her own head and it's become second-nature, she recognizes the toll it can take cognitively and emotionally for PWDs and their families -- especially when it comes to food.
"Let’s face it: without meals, we would have a fully autonomous closed-loop Artificial Pancreas today," she says. "I think Klue’s passive meal detection capabilities are a key missing piece."
What's Ahead for Klue
As Klue begins to partner with other digital health and device companies, it opens up a world of new possibilities -- especially with Closed Loop technology and some of the exciting development happening in the do-it-yourself #WeAreNotWaiting community.
While many of the current commercial and DIY closed loop systems automatically adjust basal rates and can offer boluses or tweaks based on CGM data, the PWD still needs to key in their food data in order for the system to start acting on that.
"Think about what you could do, if you got early detection when someone actually starts eating, way before the blood sugar levels start to rise," Vleugels says. "How can you merge that data point insight into your algorithms? That's something I am very excited about, and I think the potential is very big."
"We have APIs available, and we are actively reaching out to the other parties in the diabetes ecosystem. We are inviting everyone to the table, to have a discussion about how we can use the capabilities that Klue's developed to supplement, compliment, and improve the abilities that our partners have, to combine them and serve the diabetes community as a whole."
While Klue's current bolus reminder is a bit limited in functionality, we sure do see the huge potential of combining this hand-movement detection with other tools that can track carb and insulin amounts.
Here's to a bright future in the diabetes space for Klue!