Diabetes Mine 2019

Diabetes Mine 2019 Innovation Summit & D-Data Exchange November 7-8, 2019; San Francisco, CA; Highlights - Draft

Executive Highlights

  • The first ever “Closed Loop Showcase” shared presentations on seven (!) automated insulin delivery systems: Medtronic’s MiniMed 780G, Tandem’s Control-IQ, Insulet’s Horizon, Tidepool Loop, Beta Bionics iLet, Bigfoot Unity, and DIY. Highlights included (i) an explanation of how Tandem’s Control-IQ algorithm works (currently under FDA review) and a trial participant’s experience; (ii) an update on Tidepool Loop – over 874 people are in the observational study (!) and Jaeb is starting to compile data (plus two more device companies have term sheets to sign on); (iii) pre-pivotal data on the MiniMed 780G (currently in its pivotal study); (iv) a look at the Omnipod Horizon’s commercial user interface (pivotal start next month); (v) updated timing on Bigfoot's’ Unity “dose decisioning system for MDI users (2021 launch); and (vi) one study participant’s experience from Beta Bionics’ Gen 3 iLet bridging study. Wow was this an awesome look at the next-generation of automated insulin delivery!

  • During the demo portion of the day, Ascensia unveiled its diabetes coaching and bundled program. It is currently in a commercial pilot, with a full launch expected in 2020. The beautiful design (with help from IDEO) includes a smartphone app, starter kit, unlimited stirps, and coaching focused on small, behavior modifications called “bursts.” Other demos covered below include UndermyFork (fantastic CGM + meal photo app), Cam Med’s EvoPump, and Pacific Diabetes Tech’s single-port CGM/insulin catheter.

  • We also enjoyed hearing panels on “New Clinics for a New Age” (Onduo, Steady Health, Hygieia) and “The Rise of Patient Voices” (Novo Nordisk, FDA, UCSF). Of note, Novo Nordisk’s Joseph Thinggarrd highlighted the company’s smart pen efforts, which are still expected to begin launching next year – durable pens will start “scaling” in 2020, with “parallel development” of a disposable pen attachment.

Hello from UCSF’s Mission Bay Conference Center, where Diabetes Mine held its 2019 Innovation Summit & D-Data Exchange over two packed days. See our top highlights below!

All the slides from the day are posted here!

Closed-Loop Showcase Highlights

1. Tandem’s Control-IQ Algorithm Rundown, Simple Five-Step Online Pump Update Process; One Trial Participant’s Experience

Tandem Director of Product Management Garrett Marin gave a brief rundown on the highly-anticipated, under-FDA-review Control-IQ algorithm, including a useful chart depicting how the modular algorithm works. The algorithm, licensed from TypeZero/Dexcom, uses CGM readings, insulin-on-board, programmed pump settings (e.g., basal rate, insulin-to-carb ratio, and insulin correction factor), and “other variables” to predict blood glucose values 30 minutes into the future. When the predicted insulin value is between 112.5-160 mg/dl, the algorithm maintains the “active Personal Profile”; in other words, insulin is delivered as programmed by the user. When glucose is predicted to be between 70 and 112.5 mg/dl in 30 minutes, the basal insulin delivery is decreased and if glucose is predicted to be below 70 mg/dl in 30 minutes, basal insulin delivery is shut off. Similarly, basal insulin is increased when blood glucose is predicted to be between 160-180 mg/dl in 30 minutes and a correction bolus is delivered when glucose is predicted to be above 180 mg/dl in 30 minutes. These automated correction boluses are calculated as approximately 60% of a correction bolus needed to reach a target of 110 mg/dl. When Sleep Activity is activated (either by a programmed sleep schedule or manually enabled by the user), the adjustment ranges are lowered and narrows in order to achieve 112.5-120 mg/dl by the morning. Conversely, when Exercise Activity is enabled, the ranges are raised to achieve glucose values between 140 and 160 mg/dl. The chart below summarizes the algorithm very nicely! Control-IQ was submitted to the FDA in ~July and is expected to receive approval in the next six weeks, launching as a free software update. Read more below on one trial participant’s experience!

  • With the disclaimer that Control-IQ is still under FDA review, Mr. Marin gave a walkthrough of the planned five-step update process. In-warranty users will receive an email announcing availability of Control-IQ, which will take users to an “online portal” to request a prescription. When Tandem receives the signed prescription, users will be sent an online training. After completing the training and passing a “comprehensive quiz,” a unique code will be sent to users to download Control-IQ. As a reminder, Control-IQ is expected to launch in the next six weeks on new pumps and as a free software update for all in-warranty t:slim X2 users.

  • Mr. Marin provided a useful analogy to compare Control-IQ and Basal-IQ: “The math behind Basal-IQ and Control-IQ are completely different. Basal-IQ is like a light switch, in that it predicts ahead as a straight line. It takes the last four [glucose] points and, based on that, turns off the lights, or basal rate. Then, it turns it right back on as [glucose levels] start to come up. Control-IQ is entirely different mathematics. It’s more like a dimmer switch. You start to slowly turn down the dimmer switch as [glucose] goes down and as needed, we turn off. Not until you’re predicted to be back above 70 mg/dl do we start to turn the dimmer switch back up.”

Ms. Sally Southard, a participant in the Control-IQ pivotal trial (ADA 2019, NEJM) shared some of her takeaways from using Tandem/Dexcom’s advanced hybrid closed loop. Ms. Southard was actually randomized to the sensor-augmented pump control group during the trial, but has been using the Tandem t:slim X2 with Control-IQ since the extension phase of the trial.

On Using Control-IQ

  • I don’t have to think about my diabetes as much as I used to – I could cry about that. I really do trust the system, it is amazing. Because I’m not having to think about lows or highs as much, I can live my life. I truly do trust it; I’ve not had any bad experiences with it.”

  • “How hard or easy was it to learn the system? Because I’ve been on pumps so long, and I already had the Dexcom G5 CGM system, I was familiar with it. I was also a pump trainer in my role as a nurse practitioner – I was a certified trainer for Medtronic, MiniMed, Animas, lots of those – so it was easier for me to figure it out. All I had to learn were the new closed loop features, but it’s so intuitive - it’s really easy … It really is very intuitive to just look at the screens on the pump.”

On Benefits of Control-IQ

  • “Much more time-in-range. I get these [Dexcom] notifications on my phone that say, ‘You met your goal of time-in-range yesterday.’ 70% is my goal. One day, I got one that said 86% is your time-in-range, and one was 93%! That’s my high. I’m definitely spending a lot more time-in-range.”

  • “I don’t have to [give automatic correction boluses] … It’s giving me these auto-correction boluses, and I don’t have to think about it. That is huge. Normally, I’d be looking and saying, ‘It’s high, let me manually bolus.’ Then, I’d wait some, and check again, maybe do a bolus again. Now, I could be working or doing other things and not have to worry as much, because I totally trust the system.”

  • “Because of [Control-IQ’s sleep activity feature], I don’t have hypoglycemia. It isn’t perfect, it’s not a pancreas, but it’s pretty close. I still have lows and I still have highs, but I don’t have them as much as I did. My husband can tell you the alarm is not going off as much at night. I’m sleeping better and he’s sleeping better.”

  • “My [blood glucose] doesn’t drop when I’m exercising. When I do walk, there were times when I was going low and I was always took my Mentos or glucose with me. Now, I can do it without that – rarely, does it drop anymore.”

  • “Normally, when you go low, you have to eat carbs, and you can gain weight from doing that. I didn’t like that part. Now, I’m not going low as often, and when I do go low, I don’t need as many carbs. I used to take 15 grams, now I take 10 grams. I guess that’s because the pump has stopped and I’m not getting insulin.”

On Challenges with Control-IQ

  • “The problem with me, and this is true about any pump, is that … lots of women wear their pumps in their bra, or at the top of their pants, or something around their thigh if they’re wearing a dress. That’s really a pain to get to when you’re trying to bolus. Medtronic used to have a little remote-control thing I kept with my car keys, so that when I was driving long distance and wanted a snack, I could just bolus with that thing. The biggest challenge for me now is just having to remove the pump from clothing to do a bolus … Whether it’s a remote, or something on your phone, I’d like to give boluses without having to get to my pump.”

  • “The system is a little slow, or slower than I’d like, to bring down high blood sugars. Let’s say I miscalculate my carbs, which I do, obviously, then [my blood glucose] is high after a meal and I want it to come down faster.”

  • “Because it is Bluetooth, there’s sometimes loss of signal when I sleep … It doesn’t bother me at all, but parents and friends of mine have told me that sometimes children are in the trial and they see that. You know that parents are glued to their watches and looking at their children’s blood sugars.”

2. Tidepool Loop: n=874 in Observational Study, Actively Working on FDA Submission; Two More Device Makers at Term Sheet Stage; Summary of Changes from DIY Loop

Howard Look provided an exciting update on Tidepool Loop, headlined by the most detailed Jaeb observational study update we’ve seen since it began in January. An impressive 874 people are enrolled in the DIY Loop study (all ages, new and long-time Loopers; see below), and Jaeb is starting to compile the data now. Depending on the total days of use, this could very well be the largest AID data set ever compiled! Tidepool is currently working on its FDA submission, including potential for a de novo interoperable control algorithm designation (to complement iCGM and ACE Pump). The Tidepool team is actively using Dexcom’s G6 transmitter and Insulet’s Dash Omnipod Bluetooth protocols to test device communication with the Tidepool Loop iOS app. Like he did at FFL 2019, Mr. Look did not provide an exact FDA submission or launch date (“I don’t want to over-set expectations”), but it’s clear that the critical pieces are falling into place. Tidepool is continuing contract discussions with other pump and CGM makers, including two that have reached the “term sheet” stage. Potential interoperable device partners (our speculation) could include Tandem (t:slim X2 ACE Pump), Abbott (Libre 2 iCGM under FDA review), and Senseonics (iCGM filing in 2020); however, Mr. Look gave no indication of who the two companies are. (Medtronic is also a Tidepool Loop partner as of June, though its ACE Pump and iCGM are still in development.) Said Mr. Look, “This is an idea that has resonated – people love the notion of an interoperable system.”

  • “All of our interactions with the FDA are posted at tidepool.org/documents. All of our code is open source. Our program management is all out in the open. I was working on our FDA submission yesterday. Jaeb has started to pull the observational data and consolidate it. We don’t know when it will be ready – there’s too much uncertainty. But we are on it!”

    • Tidepool had a Pre-Submission Meeting with the FDA in June. The posted documents from June 2019 (mid-way down the page) are a really interesting read. Kudos to Tidepool for this unparalleled level of transparency! The June 2019 Meeting Minutes Approved by FDA provide a great 22-page summary of how FDA and Tidepool are thinking about Tidepool Loop safety, iCGM compatibility, statistical analysis, and more. Reading through this short document also shows how incredibly nuanced and involved the regulatory process is – even for a system thousands of people are already using. All the FDA interactions are publicly available at tidepool.org/documents

  • Mr. Look provided the first-ever summary of anticipated changes from DIY Loop to Tidepool Loop – e.g., safety guardrails, colors, on-boarding flow. As expected, Tidepool will add guardrails around certain settings (e.g., correction range: 60-180); the current version of DIY Loop does not have these. Tidepool is also adding an in-app tutorial and onboarding workflow, which will greatly help with initial training over the robust (but slightly overwhelming) DIY version at loopdocs.org. The team has also made some color updates to address accessibility (more blue, based on the user interface pictured below) and to offer a quick way for customer support to determine if someone is running DIY Loop vs. Tidepool Loop. The core user interface otherwise won’t change, offering the CGM trace up top, the two insulin charts below it, and a carb absorption chart below that. Tidepool is still figuring out if “override” modes will be included in the 1.0 version – these are temporary targets, a common example being women who use them during menstrual cycles (e.g., the first five days can have different insulin levels than the rest of the cycle). This feature is also good for adjusting insulin levels during any illness. It remains unclear if override targets will require a new study, but this will “definitely” be added Tidepool Loop at some point.

  • “Iteration of software is better for public health.” Tidepool plans for “ongoing and hopefully rapid iteration of Tidepool Loop – 1.0, 1.1, 1.2, 2.0…” Added Mr. Look, “We fully anticipate regular updates to the app, just like every other app in the App Store. Some changes will be released quickly, and some will need to go through an IRB-approved/IDE-approved study. But our intent is to keep iterating. If features don’t make it in to 1.0, they’ll be close after in 1.1 and 1.2.”

  • The graph below charts DIY Loop Observational Study enrollment over time, with the big jump in April driven by the addition of Omnipod DIY Loop compatibility. Cohort A (black line) are users new to DIY Loop (started using DIY Loop 7-14 days after joining the Jaeb study) and Cohort B (red line) are users that have been using DIY Loop longer than 14 days. The “best guess” now is that there are ~4,000-7,000 people using DIY Loop globally, based on the number of RileyLinks sold. Of course, the RileyLink won’t be needed for the commercial version of Tidepool Loop, as it will communicate directly from the phone to the ACE pump and iCGM.

Observational Study of DIY Loop – Enrollment by Age


# of Participants

<7 Years


7-14 Years


14-25 Years


25-50 Years


>50 Years




3. Medtronic MiniMed 780G US Pivotal “Going Quite Well”; Pre-Pivotal Results Showed 85% TIR, 99.9% Time in Closed Loop

Medtronic engineer Mr. Lou Lintereur shared an update on the MiniMed 780G Advanced Hybrid Closed Loop (AHCL) pivotal trial and showed early pre-pivotal study results (Time in Range: 85%!). The MiniMed 780G, targeted for launch by April 2020, will bring automatic correction boluses, an adjustable set point down to 100 mg/dl, Bluetooth and a mobile app, remote software updating, fewer alarms, and an ambitious >80% time-in-range goal. In a very brief update on the pivotal trial for 780G, Mr. Lintereur stated the trial is “going quite well.” Notably, on the slide, the pivotal study design is described as “single-arm, 150 patient study for ages 14+.” This is a bit different from the last update we heard in 2Q19 and the ClinicalTrials.gov page, which suggested a larger 350 patient study with patients 7+ years old. It’s possible Medtronic has now broken the pivotal into two phases, like it did for the 670G (one for Adults/Adolescents, one for Pediatrics). According to ClinicalTrials.gov, primary completion is expected by January 31, 2020. Could we see results at ATTD?

  • A slide (below) showing early feasibility study (n=12) results suggested significantly improved usability for MiniMed 780G. The average number of alarms per day was decreased from 12 in 670G to “<6” with 780G. The number of fingersticks decreased from 7.1/day to 2/day and the number of Auto Mode (i.e., closed loop) exits per week dropped from 4 to 0.3 (~once every three weeks)! Impressively, all participants reported the 780G “Advanced Hybrid Closed Loop” system reported their overall satisfaction as “extremely” or “very satisfied.” Full results from this study were published in last month’s issue of DT&T, and earlier results were shared at ADA as posters (1040-P and 1041-P). The study, which took place in June/July 2018 at one site in Australia, used an earlier prototype of 780G, called “enhanced-HCL” (different from 780G’s “Advanced HCL”). Participants went through a one-week, open-loop, at-home, run-in phase, followed by a five-day, supervised hotel stay after closed-loop initiation (to ensure safety), and then three-weeks of enhanced-HCL use at-home.

  • Time-in-range improved from 75% in the open-loop period to 85% during the at-home closed-loop period (+2.4 hours/day; p=0.003). The 10% absolute improvement in TIR is very impressive, given the already high baseline of 75%. We will be curious to see how these numbers look for the commercial product in the much larger US pivotal trial.


Open-loop (1 week)

Closed-loop (3 weeks)


Time 70-180 mg/dl




Time 70-140 mg/dl




  • Not surprisingly, the time-in-range improvement was driven by decreased time in hyperglycemia. Percent time above 180 mg/dl was halved from ~21% in open-loop to ~10% in closed-loop (p=0.002), while percent time above 250 mg/dl was reduced from ~3% to ~1% (p=0.006). 


Open-loop (1 week)

Closed-loop (3 weeks)


Time >180 mg/dl




Time >250 mg/dl




Time >300 mg/dl




  • Time below 70 mg/dl was a touch higher in the closed-loop period (3% vs. 4.4%; p=0.02), though still around the recommendation for 4% or less time <70 mg/dl. However, the periods showed no significant difference in time below 54 mg/dl (0.4% vs. 0.6%: p=0.6).


Open-loop (1 week)

Closed-loop (3 weeks)


Time <70 mg/dl




Time <54 mg/dl




  • Mean sensor glucose was 123 mg/dl during closed-loop, compared to 144 mg/dl during run-in. Coefficient of variability was not significantly different between the two periods.


Open-loop (1 week)

Closed-loop (3 weeks)


Mean sensor glucose

144 mg/dl

123 mg/dl


Coefficient of variation




  • Time spent in closed loop was an impressive 99.98%, with a median of one Auto Mode exit during the three-week closed-loop period. That is a major improvement over 87% time in closed loop seen during the pivotal trial for 670G and far more Auto Mode exits.

  • Patient-satisfaction metrics were very positive, with all patients reporting the system was the best therapy they had ever used. Seven of 12 participants rated their likelihood of recommending the system to a friend as 10/10 and eight out of 12 said the system “far exceeded expectations.”

Selected Question and Answer

Q: Is handling physical activity somewhere on your roadmap?

A: That is somewhere on our roadmap, not specifically for Advanced Hybrid Closed Loop. We are touching it a little bit. If you’re familiar with the 670G system, the fixed target is 120 mg/dl, but we also have a temporary target for up to 12 hours at 150 mg/dl. Internally, we called that “exercise mode,” because we thought that would be the most common use case. That feature remains in the Advanced Hybrid Closed Loop, but one thing that’s different is you can now set it for a full 24 hours. 

4. An Updated Look at Omnipod Horizon’s Hybrid Closed Loop Interface

Ms. Lindsay Philbrick (Insulet) shared the first public demonstration of its final interface for Insulet’s hybrid closed loop system, Omnipod Horizon. In Insulet’s 3Q19 call last week, we learned that Insulet has already completed Horizon’s formative human factors testing and is in the summative human factors phase. A three-month, 240 subject pivotal trial is expected to begin in December and US launch is still targeted for 2H20. As a reminder, Insulet plans to launch Horizon with personal smart phone control, and the demonstration was shown using an Android emulator. Previously, Insulet expected smartphone control to initially launch only on Samsung Galaxy phones, expanding to other phones over time. We’re not sure if that will change as the pivotal/submission progresses. Presumably non-Samsung-Galaxy users will be able to use Horizon with a dedicated Dash PDM.

  • The user interface is very similar to the existing interface on Dash, though Horizon will have CGM data from Dexcom on the default view, rather than insulin on board. An icon in the top right indicates whether the device is in closed loop or manual mode and an integrated button can take users straight to the Dexcom G6 app. The bolus button at the bottom of the screen includes a food library like the Dash, but also includes a “Use CGM” button to import the current CGM value for bolus calculation. Changing closed-loop modes (e.g., Hypo Protect, Temp Basal, etc.), history, and settings are all available in a hamburger menu. The Horizon interface also includes a “View” button to display CGM trace and insulin dosing data in one graph. It seems that Insulet has avoided the CGM trace on the home screen of Horizon, choosing instead to keep it simple with the CGM number and arrow.




5. Bigfoot Unity Smart Pen Launch Now “2021”; Autonomy Pump Hybrid Closed Loop Now Expected in “~2023”

In the first update since our June interview, Bigfoot CEO Jeffrey Brewer shared new timing on the company’s evolving insulin automation efforts: (i) Bigfoot Unity, a smart pen cap/FreeStyle Libre CGM “dose decisioning” system for MDI users, is now expected to launch as the first product one year later in “2021,” with a 510(k) filing expected in 2020; (ii) a next-gen version of Unity with automatic basal/bolus titration will follow in “~2022” (it will require a pivotal trial); and (iii) the pump-based Autonomy hybrid closed loop is now very distantly expected in “~2023.” We've confirmed with Bigfoot that Unity gen one will not titrate doses on its own; rather, it will display the HCP-recommended dosages on the pen cap based on current glucose and insulin sensitivity factor, and the user can decide whether to deliver that dose. In other words, Unity gen one is a precursor product ahead of the long-stated goal -"auto-titration" for MDI users (Gen two). In the second-gen version of Unity, the system will automatically adjust the dosing algorithm via personalization and individualization based upon the user's actual results using the system over time (i.e., closing the loop for injection users, with adaptation). While this certainly pushes out Bigfoot's timelines even further, we're glad to see it taking a realistic staged approach - better to get the precursor connected pen product to market and iterate from there. Mr. Brewer emphasized that Bigfoot’s initial focus is on “the largest and most underserved markets – MDI users.” We appreciated seeing an updated look at the company’s smart pen user experience, which will display insulin dosing information right on the pens – convenient and differentiated! In modeling experiments so far, the Unity system has delivered robust A1c reductions, with minimal incremental benefit to adding a pump-based closed loop on top of it. We’ll be eager to see what the pivotal trial looks like when Bigfoot gets that underway! Key slide pictures are enclosed below. Bigfoot continue to expect delivering Unity as a service – bundling all the supplies under one prescription (excluding the insulin pens). While Companion, Lilly, Novo Nordisk, and Sanofi obviously have their own smart pen efforts, Bigfoot still has a shot as stitching together a nicely integrated, compelling MDI titration system that gets close to a pump-based system at a lower cost. Of course, hitting its timelines going forward remains a big question – relative to the original ambition to be in its first pivotal trial in 2016, Bigfoot is running at least four years behind.

6. Experience with Beta Bionics’ iLet Gen 3: “The Flattest Blood Glucose Lines I’ve Experienced in my 17 Years with Type 1”

The animated Ms. Rachel Mercurio shared her incredibly positive experience moving from MDI to participate in Beta Bionics’ 2018 iLet bridging trial. The six-week trial took place in summer/fall 2018 with participants using the larger Gen 3 iLet in the insulin-only configuration. After Ms. Mercurio had the device set up, she simply entered her body weight and “let the pump take over.” It took “4-6 hours” for the iLet to warm up and adjust to the user, but after that, Ms. Mercurio described the iLet as “set it and forget it!” as the device took over “literally thousands of decisions” for her. She also noted that she really put the iLet to the test – Ms. Mercurio’s first meal after getting the iLet was pizza. With the iLet, Ms. Mercurio experienced flatter blood glucose lines than she has ever seen. Additionally, she had the best sleep she has ever had, something she called “truly a life changer.” Ms. Mercurio’s biggest complaint with iLet was that the device was bulky and heavy. The Gen 3 device used many lithium batteries every day and required an uncomfortable steel needle for insertion. However, she did note that the trial was performed with the Gen 3 iLet, and many of the issues described have already been addressed in Gen 4 iLet, which will be device used for commercialization. The Gen 4 iLet is just 15 mm thick and has a mostly icon-driven touchscreen interface that seems very user-friendly. See our Friends for Life 2019 report for pictures and a very detailed breakdown.

Taken at FFL 2019: Gen 4 iLet (left), Gen 3 iLet (middle-left), Gen 2 iLet (middle right), iPhone X (right)

  • After the iLet trial ended, Ms. Mercurio chose to go on DIY Loop. While Loop didn’t offer the same “set it and forget it” ability that she saw with iLet, Ms. Mercurio did see improvements in her A1c with Loop, compared to MDI: with Loop, she saw her A1c decrease from 7.8% to 6.7% in just four months.

  • Ms. Mercurio chose to close her presentation by reading a powerful statement on the importance of access to technology: “When it comes to this groundbreaking technology that we’ve heard about today, all the hard work, time, and effort will be for nothing, if patients can’t get access. I am grateful to have some of the best durable medical equipment coverage in the country. I have state benefits. I’m incredibly blessed. And yet, I still have a difficult time getting my supplies every 90 days. It shouldn’t be this hard. Access to these life changing products for all must happen in order to make these innovations matter.”

7. Dana Lewis on DIY: Improves Sleep, Uses “Simple” Algorithms, Quantifies Behaviors, and Illustrates Individual Variations

OpenAPS’ Dana Lewis shared her views on DIY’s appeal and ongoing themes. Download Dana’s slides here. Beyond the glycemic benefits, including reductions in A1c and improvements in Time in Range (e.g., EASD 2019, ADA 2018), she explained that DIY enhances sleep quality, employs “simple” algorithms, sheds light on how individual behaviors impact glycemic control, and illustrates individual variations. Ms. Lewis concluded that there is still much “low-hanging fruit” left to be tackled and that instead of waiting for companies, patients should be setting and moving the bar: creating a world where they get access to (i) real-time access to personal data on the device of their choice; (ii) interoperability across pump, CGM, and algorithm/controllers; (iii) more flexibility to do less work and achieve the best outcomes; (iv) and the ability to define priorities.

  • The DIWHY study conducted by the Open Consortium pooled data from patients and caregivers about motivations for DIY (34 countries, 5 continents) found a whole host of motivators for using DIY systems. We first saw data at ATTD 2019 and again in poster form at ADA 2019. We’re elated to see the DIY community continuing to validate the benefits of the technology – see an impressive list of Dana’s research here.

  • Ms. Lewis emphasized that DIY systems use simple algorithms, but they can be personalized to the user. It’s been ~27 months and counting without using manual boluses despite managing the flu, norovirus, twelve hours of jetlag, marathon training, an ankle fracture, and other daily occurrences. To the audience’s applause, Ms. Lewis also emphasized that tools like the OpenAPS Simulator can help answer algorithm questions – how does the system respond to unannounced meals without a bolus? The OpenAPS simulator was surprisingly on point in coming actual vs. simulated unannounced meals!

  • DIY technology today can also illustrate the causes of individual variations in blood sugar that doctors and people with diabetes often miss. After fracturing her ankle, Ms. Lewis described how her blood sugar immediately increased to 160 mg/dl. Approximately one week later, she tripped over her crutches and broke a toe on the other foot. Her husband, Scott, didn’t believe the toe was broken – at first. Another blood sugar climb to 160 mg/dL convinced him, and an X-Ray two days later confirmed the result. Smart algorithms, Ms. Lewis emphasized, have the potential to drive n=1 learning: does one type of insulin works better than a different one? Where are the best pump insertion locations? Where is the best body placement for CGM? What is the best target or insulin strategy for someone’s activity and exercise? Currently, these questions are either not answered at all, or they are answered with a lot of trial-and-error learning. However, the data/algorithm tech already exists to answer these types of questions based on actual data, and with much less effort.

Demo and Panel Highlights

1. Ascensia’s Coaching & BGM Bundle in Commercial Pilot Stage, Launch in 2020; Focus on Type 2 Diabetes

Highlighting the demos portion of DiabetesMine Innovation Summit 2019, Ascensia unveiled its “diabetes coaching system,” which was developed with the world-renowned design firm IDEO. It will be offered as a bundle with unlimited test strips, an app with human coaching, a Bluetooth-enabled Contour Next BGM, and lancets. Ascensia is focused on the “proactive” nature of the program, noting that its diabetes educators would look at patients’ data and reach out, rather than waiting for patients to call. Additionally, the program is focused on small behavior changes, called “bursts,” in a variety of areas: diabetes management (e.g., taking fingersticks), movement (e.g., walking), stress, sleep, and food. Participants and coaches focus on one “burst” at a time, so that the user is not overwhelmed. These bursts are designed to be quick, achievable, measurable, and jointly agreed upon by the member and their coach. We love it! The program is currently in commercial pilot with “real payers and members,” and a full commercial launch is aimed for 2020 with both direct-to-consumer and business-to-business offerings. Although the pilot version of this solution has been known as “Abound,” we Ascensia plans to launch it under a different brand name.

  • From the demo pictures, Ascensia has put together an attractive and user-friendly app interface and starter kit – not surprising, given the IDEO collaboration. Like mySugr, OneDrop (Premium), Livongo, Dario, and others. We’ll be interested to see what Ascensia’s go-to-market approach is – where will it set pricing relative to competitors? How will it balance direct-to-consumer vs. payers/employers? What about retail/brick-and-mortar (e.g., One Drop in Walmart and Apple stores) vs. in-app sales vs. ecommerce sales (e.g., Amazon).   

  • Early on in the demo, a slide briefly referenced Ascensia’s partnership with Chinese manufacturer POCTech to co-develop a CGM. Ascensia previously expected to begin selling “Ascensia branded and boxed” CGMs developed by POCTech by the end of this year, though that likely has not happened. POCTech’s CT-100B CGM system is already CE-marked and has 7-day wear, one calibration per day, a reusable battery-powered transmitter lasting up to two years, and direct Bluetooth-to-smartphone connectivity or a dedicated receiver. Ascensia still holds significant market share in the declining BGM market (~20% of global SMBG market) and adding digital coaching offerings could help make it more competitive with Roche/mySugr and others.

2. UndermyFork: CGM + Meal Photos; Insulin Logging Coming in 1Q20; Contextual Pattern Recognition (2Q20); BG Predictions in Pipeline

We were elated to see UndermyFork in its first diabetes tech conference demo – the useful app combines CGM and meal photos in an intuitive, time-in-range-focused interface. Adam first wrote about the app in diaTribe earlier this year (Bright Spots Lab #3), and both Adam and Kelly have been using it ever since. Building on the brilliant Meal Memory app from years prior, users log meals with pictures and CGM data is overlaid – including time-in-range after the meal and useful sorting (high, medium, low time-in-range meals). UndermyFork also has smart photo recognition to recognize what foods are in the meal; it’s correct in identifying foods ~70% of the time, in our experience. Users can then search for the last time they ate something. Currently the iPhone app accepts data from Dexcom CGM via Apple Health, FreeStyle Libre for those using the off-label Miao Miao/BluCon add-on, and several BGMs. UndermyFork is looking for partnerships to get tighter glucose data flow into the app via API/cloud integration – we think this would be huge value-add for any glucose monitoring device, as this app can add extremely useful meal-photo context to improve time-in-range (and much lower hassle than typing meals in via text). UndermyFork aims to support all CGMs. Near-term, the company is at work on a weekly summary report that would synthesize highest/lowest time-in-range meals – much like Dexcom’s Clarity notifications or Fitbit Progress reports. Insulin logging will be added in 1Q20 (the most requested feature), an Android app in 2Q20, contextual alerts on iPhone and Android in 3Q20 (“You tend to go low at night when your last meal was before 8pm”), and blood glucose prediction in 2020-2021 (“What will my blood sugar be if I eat an apple and take one unit of insulin”). We’re big fans of the app and hope to see many more companies supporting it.


3. Cam Med EvoPump and Pacific Diabetes Technology Combined CGM/Insulin Catheter

On the hardware side, Cam Med and Pacific Diabetes Technology described their in-development products. Each company described a particular challenge with existing diabetes technology (e.g., too much gear on the body, ability to only deliver one hormone, infusion site inflammation), presented preliminary data, and addressed how patient perspectives are being incorporated as part of the R&D process.

  • Cam Med featured its ultra-thin, flexible, bandage-like Evopump capable of delivering multiple drugs at minimum cost. On par with the ATTD 2019 exhibit hall timing, the company still aims to complete a prototype system for human use and submit for IDE in 2020. If all goes well, it will move into a clinical trial in 2021. The three-day disposable pump uses a unique reservoir array and an electrolysis-driven system. Onboarding was described as a simple four-step process: (i) fill insulin into the EvoPump; (ii) remove an adhesive release liner; (iii) place the EvoPump on the body and rotate the applicator; (iv) and pull the applicator up and off. Cam Med hopes to make the device interoperable, pending FDA submission. The company is completing a full-scale beta prototype this year and characterized patient feedback as positive. JDRF funded the company back in 2018.

  • Pacific Diabetes Technology (PDT) featured its Continuous Glucose Monitoring Infusion Port – combining an insulin catheter and CGM sensor in a single site. One slide suggested it could be combined with a smart pen, implying the company is also aiming for a CGM-only product (or perhaps a product like Medtronic’s injection port for MDI users). An initial 2018 pilot study in Oregon found that the CGM has a MARD in the 10%-14% range. We note that development has been fairly slow on this product, though the aim is ambitious! When asked about patient input during Q&A, the rep mentioned that the team has actively recruited employees with diabetes and has reached out to people who use AID and smart pen technology. PDT is currently “gathering all the input” available at this time and is trying to “get as many people in the testing phases” as it moves closer to market.

4. Connected Care Panel with Onduo, Steady Health, Hygieia; Challenges/Opportunities for Remote Diabetes Care

Harvard Professor Dr. Ateev Mehrotra led a panel featuring reps from Onduo, Hygieia, and Steady Health, highlighting the differences among these new care delivery models and challenging traditional views of the impact of technology on diabetes health outcomes. Dr. Mehrotra showed a series of slides deftly questioning a number of common assumptions about telemedicine: Does it decrease cost? Does it decrease health disparities and expand access to rural communities? Does it demonstrate the same quality of care compared to in-person PCP visits? Does it reduce utilization? In each case, there was data to suggest the opposite of what is hoped for. For example, in the direct-to-consumer space, new care options (e.g., MinuteClinics) actually appear to increase disparities – almost 90% of retail clinics are in urban areas, while only 13% are in health professional shortage areas (vs. 21% of the population). With respect to quality, he cited frequent reports of inappropriate use of antibiotics and testing. Furthermore, while new technology can be cheaper, healthcare spending does not necessarily decrease , especially if there is new utilization driven by convenience. For example, 65% of retail clinics and 88% of direct-to-consumer (DTC) telemedicine visits represent “new utilization” – i.e., they are not substituting for in-person visits, but adding to it. Dr. Mehrotra did a masterful job laying out some of this idea, and then asking each connected care company a series of curious questions.


  • Onduo’s Patrick Cunningham displayed a compelling map showing how the company’s “virtual diabetes clinic” (connected Telcare BGM/Dexcom CGM + telemedicine coaching + app) has scaled access to endocrinologists in regions where few or none exist. He showed data up to six months highlighting average A1c reductions of 2.3% for those with a baseline >9.0%; a 0.7% reduction from a baseline of 8.0%-9.0%; and a 0.2% reduction from a baseline of 7.0%-8.0% (abstract here). Onduo is now operational in 49 states, requiring quite a significant investment to get licensed medical doctors across the US. Onduo has been expanding into commercialization this year, and initiatives in the pipeline include: (i) a randomized controlled trial; (ii) intermittent use of real-time CGM; (iii) medication adherence and optimization; (iv) engagement and lifestyle interventions. It’s unclear the exact number of patients who have access to the platform; see our last coverage of Onduo from ATTD 2019.

  • Steady Health is a new full-scope diabetes clinic in San Francisco that connects patients with CGM and pairs them with a clinical team consisting of an endocrinologist, a Care Coordinator, and CDE. Similar to One Medical or other concierge clinics, Steady has built its own system of care, strong emphasis on modern tech, and a cozy in-person office in downtown SF. Contextual data is collected via Steady’s mobile app (Apple, Android), where users can chat with the care team. The approach centers around using CGM to design goals diet (photo-based logging), medication, and exercise. Steady’s team includes an on-staff endo, Dr. Calvin Wu, and respected diabetes educator Aimee Jose. The program requires one face-to-face interaction per year and offers remote coaching done exclusively through in-app text chat and video. Steady charges $59/month fee, plus in-person visit cost. Steady is already in-network for "most" private preferred providers organizations and the cost of the once per year in-person visit is billed as a regular office visit.  Thus far, the team has received a lot of referrals from providers in San Francisco. The idea here is in many ways similar to Diabeter (acquired in 2014 by Medtronic, though without recent updates). While no initial data on the program was provided, it adds a nice in-person, modern clinic approach to the exciting list of connected care initiatives. See our deeper dive on Steady Health from ADA 2019 and the clinic’s informative website.

  • Hygieia CEO Dr. Eran Bashan presented on the d-Nav Insulin Guidance Service, currently available in the UK and US. The insulin titration service is focused on injection users, combining a care center and the d-Nav device (BGM with built-in titration algorithm) to optimize insulin therapy. To date, Hygieia has prescribed a remarkable 3+ million doses of insulin, and its study with BCBS Michigan showed nice cost savings. According to Dr. Bashan, 90% of users improve within three months. As of last fall, the d-Nav Service became available to all BCBSM members in Southeast Michigan who use insulin to treat their type 2 and had an A1c >7%. (Hygieia has opened its first care centers in Southeast Michigan.) The agreement followed a direct BCBSM investment in Hygieia to support the cost of a 1,000-patient, 12-month reimbursement study. At ADA, a poster presented extremely compelling nine-month data from the demo project: A1c at the nine-month mark dropped a remarkable 2.3% (from a high baseline of 9.4%).

5. The Rise of Patient Voices – FDA, Novo Nordisk, and UCSF on Impact in R&D, Clinical Settings, and Public Policy

A powerhouse panel with the FDA’s Michelle Tarver, Novo Nordisk’s Jonas Thinggaard, and UCSF’s Susan Pappas discussed integrating patient voices into their initiatives. Mr. Thinggaard described how Novo Nordisk employs a team of anthropologists, user researchers, and systems engineers early in the product development lifecycle who conduct ethnographic research to inform what products and solutions to develop. The company engages in an iterative process, which tests very early prototypes with users and pools user feedback to refine the final product, understand user preferences and value proposition, and identify emergent safety risks. Mr. Thinggaard directly acknowledged that this process has informed development of the NovoPen 6 and Echo Plus, which will being to scale in 2020. In parallel, the prefilled add-on will be developed, implying we may not see it until much later in 2020 or perhaps in 2021.

  • Ms. Pappas mentioned that UCSF has 16 Patient & Family Advisory Councils that conduct patient panels, engage focus groups, and review policy. During Q&A, she described how UCSF had battles with some diabetes patients who wanted to self-administer insulin in the hospital. Through qualitative data collection and patient engagement about patient priorities, UCSF adopted a new policy allowing “opt-out” –  it has led to a dramatic drop in complaints. She also announced plans for additional patient focus groups in the pipeline and described the immense progress the US has made on patient engagement since the 1980s. At that time, small surveys would go out for delivery to patients, but often ended up on shelves.

  • Dr. Tarver described how the FDA created caregiver pilot organizations which pooled data from over 4,000 patients on their experiences with medical devices. The agency also developed the Patient Engagement Advisory Committee (PEAC), which provides recommendations about issues that impact medical devices. It is the only advisory committee whose members are all patients or advocates. Members complete and submit “homework assignments” that generate feedback and attend 1-2 meetings per year (see our coverage of September 2019’s meeting on cybersecurity). Overall, Dr. Tarver emphasized that gathering patient input at the FDA is a scientifically-driven process and the FDA’s commitment to incorporating these perspectives in policy.


--by Anirudh Gururaj, Albert Cai, Adam Brown, and Kelly Close