Medtronic came out with two pieces of news yesterday around its connected diabetes efforts: an FDA clearance for a hybrid closed loop insulin delivery system and the first test cohort of its Sugar.IQ app, created in collaboration with IBM Watson.

The FDA approval of Medtronic’s MiniMed 670G hybrid closed looped system was announced by the FDA itself, as it’s the first automated insulin delivery system to get clearance by the agency.

“The FDA is dedicated to making technologies available that can help improve the quality of life for those with chronic diseases – especially those that require day-to-day maintenance and ongoing attention,” Dr. Jeffrey Shuren, director of the FDA’s Center for Devices and Radiological Health, said in a statement. “This first-of-its-kind technology can provide people with type 1 diabetes greater freedom to live their lives without having to consistently and manually monitor baseline glucose levels and administer insulin.”

The clearance for the system, which uses a Medtronic Guardian CGM and a new algorithm called SmartGuard, arrived sooner than most in the industry expected.

""The FDA approval of the world's first hybrid closed loop system is a culmination of many years of hard work and close collaboration with the clinical and patient communities to generate the body of evidence needed to advance this technology for those living with diabetes," Dr. Francine Kaufman, chief medical officer of the Diabetes Group at Medtronic, said in a statement. "We appreciate the unprecedented speed by which the agency approved our PMA submission to help bring this advanced insulin pump therapy so quickly to US patients living with this challenging disease. We are committed to preparing for commercial launch as quickly as possible while ensuring we provide the most successful rollout of this novel therapy."

Closed loop or “artificial pancreas” systems combine a CGM and an insulin pump to regulate a user’s insulin with minimal interaction required from the patient. The system isn’t completely automated — users need manually request insulin doses to counter carbohydrate (meal) consumption.

Interestingly, Sugar.IQ, which was shown off at Health 2.0 in Santa Clara this week, could help with that piece of the puzzle. Originally announced at CES, the app is now a more or less finished product and will shortly began testing in its first 100 or so users, people with diabetes out in the world, not in a clinic setting. 

The app includes a manual food log and integrates data from Medtronic MiniMed Connect and the soon-to-be-released Guardian Connect app. As users record data about what they eat, when they use insulin, and their blood glucose levels, Watson machine learning generates predictive insights. So if a user enters tuna salad in the log, the app might respond with an observation that four of the last five times they ate tuna salad, their blood glucose ended up going low.

“The one question a person with diabetes is trying to answer is ‘How much insulin do I need right now?’ That’s the one question people have,” Huzefa Neemuchwala, senior director of data and informatics innovation at Medtronic, told MobiHealthNews on the sidelines of Health 2.0. “And that’s the one question the FDA will not let us directly answer because it depends on so many things. How did you sleep last night? Are you sick? Are you stressed out this morning? You know. What is your blood sugar? What is your historic BS? There are so many factors that affect how much insulin you take right now. But if you bring in more and more of these data streams, you can get at that answer by saying ‘in a similar circumstance in the past, you took this much insulin, and this was your response. ‘That’s the closest you can get to a real answer.”

Medtronic and IBM are also finetuning how and when the app delivers its insights to make sure they have the intended effects on user behavior.

“A lot of this has to do with giving you this insight when it really matters to you,” Neemuchwala said. “There’s a lot of curation and that was probably the hardest part of the project. You don’t know what Watson’s going to discover, there’s all these insights that are possible, and then curating those insights because I don’t want to overwhelm you either. I need to give you just enough so you can digest it and then take you to the next level, and hopefully that will create some of the behavior change that we aim for.”

Sugar.IQ delivers insights, not recommendations or prescriptions, so it’s in a much lower risk category than the closed-loop system, but it will still require FDA clearance before the app launches, something the company is planning within a year. 

The app will continue to be improved and iterated after it launches, Neemuchwala said. Plans include an Android version (currently it’s only iOS); integration of other data like activity and sleep, likely sourced from other digital health devices like fitness trackers; more sophisticated multivariant insights; and the ability for users to ask Watson questions and get particular insights on demand.