The Inside Scoop on the Artificial Pancreas


*Editor’s Note: This article was originally published in response to the 670G receiving FDA approval. As of May 2017, it is available commercially in the United States. 

artificial-pancreas-2The first artificial pancreas, Medtronic’s MiniMed 670G hybrid closed-loop system, has been FDA approved—a huge technological advancement in care of type 1 diabetes (an auto-immune disease). Beyond Type 1 talked to two young adults with type 1 who have already tested the product in the clinical trial of October 2015 – October 2016 to find out what kind of benefits they saw, if any, with this new technology.

The device is not fully automatic or completely “close-looped” though, as those with type 1 diabetes will still have to count carbs, input the information as well as bolus for meals or snacks. The big difference with this device being that it administers micro-boluses for basal-rate modulation on its own throughout the day and night. That means when the machine is in auto-mode, it acts like a pancreas might, secreting small doses of insulin into the body based on continuous glucose monitor (CGM) readings and without user adjustment. Effectively, the individual then can keep blood sugar levels in a healthier range without constantly adjusting manually.  Especially with a chronic illness that requires continual self-care, this technology would alleviate some of that care burden as well as administer more accurate dosing of insulin.

Using the existing technology of a pump and CGM, the system implements an algorithm that calculates insulin dosing 24/7 based on user input and the history of the user’s blood glucose levels. Studies showed that users of the system were able to stay in desired blood-glucose range 72 percent of the time, whereas left to their own dosing discretion, were only able to meet target ranges 67 percent of the time. Staying closer to a target blood glucose means less risk for debilitating and sometimes life-threatening diabetes complications that come with high levels of sugar in the blood. This is what participants of the trial had to say about their own results.

BT1: What were the requirements to participate in the study?

Participant 1: Basically, you had to be healthy (A1c less than 10.0, no severe hypoglycemic incidences, no diabetic ketoacidosis (DKA) in the past six months). Because this was an insulin pump study, you had to have been using an insulin pump for at least six months. Other criteria were: willing to check blood glucose (BG) at least four times a day, uploading weekly and participating in the hotel study. You also had to have had type 1 diabetes for at least a year.

BT1: How long have you had type 1 diabetes?

Participant 1: Almost four years.

Participant 2: 11 years.

BT1: What motivated you to participate in the study?

Participant 1: Ever since starting college, I’ve been super interested in research and being able to participate in a pivotal study was an amazing opportunity.

Participant 2: I had wanted to do a study and researched “artificial pancreas” on the web. I saw something about it, so called Stanford to see if I could be a part of the trial. I’ve found that participating in studies add a variety to my care and have helped me to stay motivated, working through cycles of diabetes burnout.

BT1: Did you have any fears going in to the study?
Participant 1: I actually didn’t. A lot of people were skeptical about having an algorithm having some control over their diabetes care. Because of the nature of the pump, there are a lot of safety features and I knew that there would be a medical team ready to help if there were any problems.

Participant 2: I knew a lot of people had trouble trusting a machine to take care of them. I didn’t feel that way, because I knew we were in good hands with the daily check-ins and excellent staff at Stanford. I knew that they wouldn’t put anyone in the position to risk something bad happening.

BT1: What was it like to give over some control of your type 1 diabetes (T1D) management to a machine in this way (especially through the night)? 
Participant 1: The pump initially has a 48-hour warm up period where it “learns” about your insulin needs. After two days, the pump will go into HCL and it should know your average delivery. Letting the pump/CGM/algorithm work overnight, I was a little skeptical at first but the CGM alarms if you’re high or low. When everything is working, it’s nice to not have to worry so much.
Participant 2: Sometimes you were high and you wanted to correct, but the machine said, “No bolus recommended,” because you were being given micro-boluses and so needed to wait to go down. That was weird to get used to, waiting for the machine to correct it.
BT1: How often did you meet with medical staff?
Participant 1: Throughout the entire study, MANY times — especially during the training period. I didn’t have any formal CGM training so I had to contact them for help in the beginning. After the initial training and finally becoming comfortable, about once every 3 months to check in and get supplies.
Participant 2: At first on a weekly basis, daily uploads and daily phone calls for the first two to three weeks to a month. Because it was a whole new pump, we had a five-hour training session. We had to wear the pump and sensor separate (no closed-loop system at first). Devices had to “learn” our blood sugar tendencies. If you went near the closed-loop system, you’d be kicked out of study.
What did you find most challenging throughout the study?
Participant 1: I think the most challenging part was switching to a different insulin pump. I was originally on the t:slim and I was really used to the touch screen and the ability to charge your pump whenever. I wasn’t on a CGM prior (although I have used one in the past), so having alarms was an adjustment.
I feel like with anything, getting used to a new system can be kind of frustrating. I knew how a CGM worked, but using the Enlite is a little different than the CGM I used in the past. The pump and CGM work together with the algorithm, so if the CGM is being iffy (no CGM is 100 percent accurate all the time) the safety features would kick in and get a little annoying.
Participant 2: There were a lot of alarms and involvement from patient. I worked the same amount of time I would with my pump and CGM. It still required lots of calibrations, lots of user input. If every little screen wasn’t accurate it wouldn’t work; it would kick you out.
BT1: Did you see any benefits from using the hybrid closed-loop system? If so, what were they?
Participant 1: This pump isn’t a fully closed loop system; it’s a hybrid closed loop (HCL) system, meaning that it modulates your basal rate through micro-boluses. It’s useful because it mimics the pancreas in the sense that your pancreas secretes small amounts of insulin throughout the day. So there’s a lot less stress about figuring out your basal rate. But it’s only meant for basal modulation and can’t account for meals or snacks. You still have to bolus.

Participant 2: Before I just had a pump and CGM, so I was rocking the midnight and 3 a.m. check. With this system, I didn’t have to do that anymore. I have a hard time with dawn phenomenon, too. And it really helped. I got way better results.

Did you notice improved BG through the night? 
Participant 1: I personally did. At the start of the trial, my A1c was 5.3 and it wasn’t due to excellent control. I am hypoglycemia unaware and my endocrinologist thought it was due to nighttime lows (I was not on a CGM at the time). With this pump, I wake up between 110 and 140 but I can see when the pump stops delivery and prevents me from going low.
Overall, this system is great for basal modulation. Having the suspend feature and CGM combined into one system has helped me prevent many lows. I don’t go low that often anymore. This system doesn’t act fast enough for miscalculated carbs at meals. It’s still best to pre-bolus and try your best at carb counting.
Participant 2: 90 percent of the time I woke up between 5.0-6.7 mmol/L90-120 mg/dL.
BT1: How was your quality of sleep affected?
Participant 1: This system has a lot of alarms. It’s part of the safety features. Unfortunately, I don’t need much insulin at night and this system will alarm when you receive the minimum amount of insulin (required by the algorithm) after two hours. Because of my nighttime insulin needs are decreased, the pump alarms probably two to three nights per week. It’s annoying, but it’s for safety.

Participant 2: I woke up feeling better. I had more energy throughout the day and was in a better mood.

Was this consistent throughout the trial?

Participant 2: Yes.

BT1: What did you think of the pump quality?
Participant 1: Pump quality … I think that because of the variety of pumps on the market, quality can be defined differently from person to person. As I mentioned before, I came from a quiet touch screen insulin pump. I had to get used to/am still getting used to the physical buttons. I’m not a huge fan of the button clicking when I’m in class. It makes it a little harder to be discrete. I am glad to hear that this pump has joined the market as a “waterproof” pump. I really like the integrated pump clip (I think Animas also has an integrated pump clip, but my sad t:slim does not). From what I hear, this pump clip is way better (and stronger) than the previous pump clip for Medtronic pumps. The screen looks really nice (a huge upgrade from the calculator screen from the 530g).
Participant 2: The pump can be frustrating to use because of so many alarms. I found that the pump was not sturdy either. I broke five pumps in the year of the trial. I bumped it, and it broke or had software issues, battery problems, blue tooth malfunctions.
BT1: Did you ever have issues with sensor insertion?
Participant 1: Medtronic changed the design of the inserter so it’s different from the previous generation. Since it’s an automatic insertion, I occasionally get that nervous hesitation of inserting the sensor (this also happens when I use the Mios/Insets). But overall, the insertion is pretty straightforward. 
Participant 2: The sensor made me bleed, so I had to dispose of many of them and try again. That wasn’t fun.
BT1: Would you recommend the system to others?
Participant 1: I have seen some people in the Beyond Type 1 community with fantastic control. The pump has a set target value at 6.7 mmol/L120 mg/dL. I believe with this generation, users do not have the ability to change that target value. If you have great control and aim for a different target value, this system might not be the greatest (you also have the option to not use the HCL and just use it as an integrated pump/CGM system like the other ones on the market). I think another criteria is using at least 8 units of insulin per day (based on what I’ve been told).
Participant 2: Definitely. The blood sugars I got with the system I’ve never seen in my life. My A1C was around 8 going into the study. After four months in the study, it was 6.1.

It felt like you got a break from diabetes in some ways because this whole staff was looking after you. There is no better feeling knowing you put in the work and got the results you wanted out of it.

*Although just FDA approved, Medtronic’s MiniMed 670G hybrid closed-loop system won’t be available to consumers until this spring.


Read Stanford’s Bionic Pancreas Study A Patient’s Point of View by Garrett Jensen.

WRITTEN BY Michelle Boise, POSTED 10/13/16, UPDATED 10/03/22

Michelle has a BA in English literature and Spanish and a MFA in writing from the University of San Francisco. As a writer, editor and content guru, she's worked on both literary magazines and e-commerce platforms. Before joining the Beyond Type 1 team, she developed health-conscious articles for Fitbit. When she's not writing, you can find her cooking or painting with acrylics (they dry faster). She is an expert decipher in the very serious discipline of finding hidden meaning in the seemingly meaningless.