November 1-2, 2018; San Francisco, CA; Day #1-2 Highlights - Draft

Greetings from UCSF’s Mission Bay Conference Center, home of the 2018 university themed DiabetesMine Innovation Summit and #DData ExChange. As always, DiabetesMine founder Ms. Amy Tenderich compiled a killer agenda, bringing together high-level market perspectives, a handful of exciting startups, and big diabetes players for annual updates.

Many of the slide decks are posted here. Read on for our highlights!

Top Six Highlights

1. Excellent Panel with Leaders from BCG, Onduo, Verily, Fitbit, Samsung; Onduo Launching Broadly in 2019; ~2% A1c Drops often seen in People with Baseline >9% (some -3% A1c drops in those using on CGM “sprints”)

Representatives from Onduo, Verily, Fitbit, and Samsung sat on a fascinating panel led by BCG and shared company updates and interesting tidbits. The major news is that Onduo intends to scale broadly next year (closing the “pilot” chapter), but we also heard that Verily’s retinopathy workflow experimentation in India continues, got some insight into why Twine Health sought to be acquired by a company like Fitbit, and Samsung’s Mr. Kevin Jones positioned diabetes as the “tip of the spear” of the company’s digital health initiatives.

• Onduo Head of Payer Relations Ms. Susanna Visuri Monacelli announced that Onduo will wrap up its commercial pilot by the end of the year, aiming to go “full scale” with payers, employers, and health systems nationally in 2019. The Sanofi-Verily joint venture commenced a small BCBS pilot in Georgia, Arkansas, and South Carolina in February and announced an expansion with BCBS Georgia in September, making the Onduo “virtual diabetes clinic” available to up to ~75,000 type 2 patients in the state. At a recent Politico event, CEO Dr. Josh Riff said that ~2,000 individuals had been enrolled to date, and touted A1c reductions in those who came in with higher baselines. Ms. Monacelli offered more granularity: Participants who had A1c >9% regularly saw 2%+ drops in A1c (the average was not shared), and those who had high baseline A1c and used real-time CGM often saw 3%+ A1c drops! Ms. Monacelli noted that CGM is being used in intermittent “sprints,” which help people make connections between their activity, food, and glucose – we love this “sprint” idea, which seeks to drive behavior change with real-time CGM, but without 24/7 use. Unsurprisingly, ~70% of people enrolled so far prefer to text with their coaches rather than talk on the phone. We also learned that: (i) Onduo aims to bill through a claims-based system, delivering care to people in “a very affordable way” (there were no copays in the BCBS pilot, and Ms. Monacelli said billing for strips and other incremental pieces “causes 100 paper cuts along the way”); and (ii) a number of clinicians have called Onduo’s office to ask how they can get the rest of their patients enrolled. We can see how an overwhelmed clinician would want more of his/her patients in Onduo’s program: the in-house care team calls the patients’ HCPs if needed, offers telehealth, and sends care summary reports with consolidated pharmacy information to providers. Said Ms. Monacelli, “Our goal is that when the patient goes back to the physician, they can say, ‘Hey, your diabetes is great, now let’s see what we can do for your other chronic conditions.’” We’ve long wondered how providers will respond to a third-party taking on patient care, so we’re pleased to hear this very positive reception from the pilot phase.

  • BCG’s Mr. Chris Bergstrom praised Onduo for its methodical approach to scale: “Unlike other startups, you have a reasonable bank account, so you could’ve started throwing money into the market. Your methodology, making sure that what you have works really well, starting with five people, then 10 people, then 2,000 people, 75,000 people, one million people…it’s very impressive, and a good lesson for other startups out there, even those who do have money.” This reminds us of Airbnb and many other tech companies, which did not hit scale off the bat – they started with a small group and built over time.

• Verily Director of Medical Devices Dr. Jordi Parramon confirmed that the company continues to experiment with deployment models for its machine-learning retinopathy detection system in India, where ~half of patients suffer from some form of vision loss at diagnosis. Verily Diabetes Clinical Lead Dr. Howard Zisser gave a deep dive on the algorithm and problem at DTM 2017, sharing that India alone has a shortage of 127,000 (!) eye doctors. Sadly there will be no such update at DTM 2018, but we’ll hopefully hear about lessons learned to date sooner rather than later. Dr. Parramon notably expressed the hope – unclear if his own or Alphabet’s – that the balance of Alphabet revenue will shift from advertising to healthcare in the future. We, too, would love to see this shift! 

• Fitbit Medical Director Dr. John Moore – presenting one day after a modestly positive 3Q18 financial update – emphasized Fitbit’s focus on wellness and social connectedness. As former CEO of Twine Health (coaching platform acquired by Fitbit in February), Dr. Moore said that the acquisition was appealing because it allowed him to merge his clinically-robust platform with a mass appeal brand (“the consumer appeal created a gravitational pull…a lot of relationships came to us because people respected and liked Fitbit’s brand, so consumers were coming to [employers] saying ‘we want this as part of our workplace or plan’”).

• Samsung Sr. Director of Digital Health Mr. Kevin Jones called diabetes the “tip of the spear” for the company’s digital health efforts. Notably, WellDoc and Samsung launched a 12-week Diabetes Wellness Program in US back in March and the robust Samsung Health app (including American Well telehealth integration) is now available in >150 countries. Mr. Jones reminded attendees that Android holds 80% of global smartphone market share). Samsung runs one of the largest hospitals in the world, Seoul’s Samsung Medical Center, which sees >8,000 outpatients per day and employs 3,000 clinicians. We were interested to learn that the Center is a “test bed” for Samsung to experiment with various solutions and take the ones that improve outcomes to the US and other countries.

  • BCG’s Mr. Chris Bergstrom advised larger companies interested in working with startups to take a page from Samsung’s book with regards to partnership (Mr. Bergstrom was previously Chief Commercial Officer at Welldoc). While Mr. Bergstrom used to get a call once a week from bigger companies asking to learn more about Welldoc, Samsung “showed up at my door and said, ‘We are so excited about Welldoc and they told me why they know more about my company than I did. It was the most respectful thing I have ever seen, and low and behold, they became great partners and investors.”

• According to Mr. Bergstrom, the Bill and Melinda Gates Foundation, a BCG client, is reportedly exploring possibilities including sending drones with medications, dropping off robots, and having the robots perform remote surgery. Even thinking inside-the-box, imagine what this could do for the supply chain of diabetes medication and supplies!

2. D-Data Demos Headlined by Klue, Apple Watch gesture-sensing for automatic meal detection – big win for full closed loop?; Diabeloop, Dance Biopharm, DiaBits, SugarMate, bgAWARE

Once again, DiabetesMine’s D-Data demos did not disappoint, offering a slew of interesting data approaches for Apple Watch, automated insulin delivery, novel CGM displays and alarms, and a new inhaled insulin. See highlights from each company below! Klue was probably the biggest highlight, especially given Dr. Bruce Buckingham’s enthusiasm.

Klue

• Klue, an Apple Watch app for automatically detecting meal start and eating speed without user intervention, announced an awesome new diabetes-focused feature: an automatic bolus reminder. Klue uses the motion sensors on the Apple Watch and artificial intelligence to detect when someone starts eating and how fast (via gesture recognition). CEO Katelijn Vleugels has type 1 diabetes and announced the really cool new upcoming feature that will recognize when a meal is starting, pulling up the Apple Watch notification pictured below: “Eating?” “Yes, bolused”; “Snooze”; “Hypo.” Dr. Bruce Buckingham came on stage after the pitch with tremendous enthusiasm – he has used the Klue app on his Apple Watch over the past week and it has “really captured almost all” of his meals. He exclaimed that Klue “is the real key for full closed loop,” as the Watch app could detect a meal and inform an algorithm to ramp insulin accordingly – all without manual meal announcement. UVA’s highly-respected Dr. Marc Breton was also quoted in the press release, “Automated real-time information of a patient’s eating behavior, coupled with novel control algorithms, may very well enable a true ‘set and forget’ closed loop insulin delivery system. This would be a game changer…” A small clinical trial is already underway at Stanford, and the company aims to bring the app to more wearables beyond the Apple Watch over time (e.g., Android, Fitbit). The current version of Klue on the Apple app store is positioned for meal and water tracking, and we especially love that it gives real-time nudges to slow down eating – clear potential for weight loss and slowing down food consumption to give insulin boluses time to catch up. A beta sign-up page is here for the new bolus reminder feature, which is positioned for type 1 but obviously could help those with type 2 on insulin. We’re quite excited about the potential for fully closed loop with no meal announcement, along with passive meal tracking and coaching for more mindful eating. We look forward to trying Klue! The company has APIs and a partnership form here.

Diabeloop

• Diabeloop CEO Erik Huneker announced that the hybrid closed loop system is launching in France “really soon” using the Kaleido patch pump, Diabeloop’s locked-down Android controller, and Dexcom CGM. He did not give an update on CE Mark status, but did clarify it is a staged process in Europe: each of the individual AID components has to have a CE Mark, and then there is another registration filing for the full integrated system – in a way, similar to the US vision of the iPump/iCGM/iAlgorithm ecosystem. Assuming the launch happens in France before the end of the year, Diabeloop will have hit the goal timing shared at ADA. End-of-2018 launches in the Netherlands and Sweden were not mentioned, and consequently, we’d be surprised if they happened in the next 56 days. Based on the Dexcom transmitter outline (see below), it looks the initial Diabeloop launch will be with the G5. Remarkably, the Diabeloop team is ~55 people, the algorithm has been tested for over 6,000 patient-days and presented in 34 orals/abstracts/papers (most recently at ADA), and ~75% of users stop counting carbs with the system. Users can customize the algorithm’s aggressiveness – a big advantage over 670G – though we’re not clear within what bounds. Diabeloop continues to work with Cellnovo, but is initially launching with the Kaleido pump; both pump companies are obviously operating at a small scale right now, and in our view, this is the largest commercial risk for Diabeloop. Mr. Huneker said the Kaleido patch pump has ~500 users now in Europe, the first update since EASD 2017. Assuming it launches soon, Diabeloop will be second to market with hybrid closed loop in Europe, following the ongoing rollout of the MiniMed 670G. Tandem’s PLGS-only Basal-IQ with Dexcom G6 is targeted for an early 2019 launch (per September’s analyst meeting), with Control-IQ likely a 2020 launch.

  • Regarding reimbursement, Mr. Huneker noted that France focuses on paying for everything to manage a condition, implying Diabeloop would be compensated for all the medical devices to care for someone with type 1 diabetes. In other countries, the traditional component-based reimbursement model – e.g., separated payment for pump and CGM – is still more common.

Dance Biopharm

• Dance Biopharm’s Dr. John Patton shared an update on the company’s inhaled (liquid) insulin and connected inhaler: (i) phase 3 trials are now expected to start at the end of 2019, four years later than originally expected; (ii) new CEO Anne Whitaker was just brought on, bringing valuable commercial diabetes experience from Sanofi North America (see press release); and (iii) Dance aims to go public at the end of next year. Dance initially filed for an IPO in April 2014, though withdrew the filing that October, partially because of the market’s uncertainty around MannKind’s Afrezza. This presentation emphasized Dance’s “pure,” “preservative free,” “no cough,” dose-adjustable inhaled insulin formulation delivered in a connected/electronic inhaler, subtly differentiating from MannKind’s Afrezza. Mr. Patton said the insulin has a slightly faster onset than subcutaneous injected mealtime insulin (though slower than Afrezza), with a longer tail that provides some “basal” coverage too. We last heard this PK/PD update at DTM 2017, when we expressed surprise that this product was still in development following Afrezza’s challenges and insulin pricing headwinds. To its credit, Dance has raised $71 million to date – including $24.5 million in September – and will obviously need quite a significant fundraise to bring this long-in-development (8+ years) product into phase 3 and over the finish line. Dance has a joint venture with Dongbao, a Chinese insulin manufacturer, to supply the recombinant human insulin for its connected inhaler. The inhaler is supplied by Philips Medisize, per the agreement announced last December. We hadn’t realized the Dance inhaler requires use of an eye dropper to load each dose, a clunkier process than we realized. There remains a lot to prove here – fundraising, safety/efficacy in phase 3, commercial viability – though we’re glad to see the focus on connectivity and non-needle-based insulin delivery.

SugarMate

• SugarMate is a companion app for Dexcom CGM, and soon, FreeStyle Libre. The app is currently available on Apple iOS, on the web, and as a Mac app (very cool!), offering a slew of CGM enhancements not available in the current Dexcom app. An automatic phone call feature drew many “wow” responses from attendees – users can configure the app to call their phone (or a loved ones’) in cases of low blood sugar (see picture below). The app also has Alexa/Siri integration (“Siri: What’s my sugar?”), an infinite scroll graph, near-real-time Apple Watch face complication updating (Dexcom’s app is a bit slow), and customizable “tiles” showing different CGM stats – see a full list of features on the company’s website. SugarMate currently pulls CGM data in near-real-time from Dexcom Share and Nightscout; the company does not have a partnership with Dexcom. The app has a remarkable 1,858 reviews on Apple iOS (4.8/5 stars), indicating tremendous enthusiasm for its user experience. For comparison, Dexcom’s G5 app has half as many reviews and harsher grading (n=948, 3.1/5 stars), and the G6 app is even lower (n=269, 2.8/5 stars). SugarMate is currently free, though we imagine some kind of subscription/paid model will be added over time.

DiaBits

• DiaBits, another CGM companion app, offers forward-looking CGM predictions (60 minutes ahead) based on Dexcom CGM data, Apple Watch activity data, and entered food/insulin data. The app is available on Apple iOS for now, with 5,000+ users at this stage. According to a poster presented at ADA, DiaBits’ machine learning algorithm showed decent accuracy for predicting glucose values 30-60 minutes ahead of time – 83% in CEG Zone A for a 30-minute prediction and 64% in CEG Zone A for 60-minute prediction. The approach is quite similar to Loop’s glucose prediction, and therefore, we’d guess this commercially available app would actually need FDA approval. For now this DIY app seems pretty under the radar. The app is free for a month, at which point it requires a paid subscription ($8+ per month). Get more on the company’s website. Like SugarMate, DiaBits does not have a Dexcom partnership for real-time data; it pulls data from a Dexcom Share account or a Nightscout URL.

bgAWARE

• bgAWARE was the earliest stage app, bringing a unique approach to CGM alarms: using vibrations to enable “glucose awareness without disruption.” A co-winner of Stanford’s Disrupt Diabetes conference (at the time, named “Infinite Pancreas”), bgAWARE will offer unique vibrations on wearables, which map to different CGM alarms. According to presenter Matt Lumpkin, the brain quickly learns to map the vibration pattern to a CGM alarm (“sensory augmentation”), meaning a user could know they are high or low with just haptic feedback alone. The hope is to decrease time spent looking at CGM data on screens – very ambitious. The company will initially launch on Fitbit and Apple Watch, but aims to develop custom hardware with a broader menu of vibration options. In early feasibility tests, the quality of life impact is “huge,” according to Mr. Lumpkin.

3. Insulet First Pump Partner for Tidepool’s Loop, with “Many, Many More” Potential Partners in Talks; Mr. Look’s Definition of Success for Launching Loop

Less than a day after it was announced on Insulet’s 3Q18 call, Tidepool CEO Mr. Howard Look reiterated that Omnipod is the first pump company that plans to integrate with Tidepool Loop. Read our coverage from last week. Mr. Look added that his organization is working with “many, many more” pump manufacturers and he hopes to have more exciting announcements; we’d guess Tandem is the next obvious candidate, and hopefully  any company that aims to automate insulin delivery will plan to integrate! Tidepool and FDA have been using Loop as a way to pressure test the Pre-Cert working model. “During four straight days of audits, we used Loop as an example for almost everything – how will that work, and how will we show safety and efficacy under the Pre-Cert program?” An app that automates the delivery of a potentially lethal drug is without a doubt a pressure test for a novel firm-based regulatory paradigm, and we wonder what was learned. Mr. Look also previously told us the team is “considering” an iAlgorithm/iController de novo application for Loop. In the brief Q&A, Dexcom’s Mr. Chris Hanneman asked Mr. Look to define success for Tidepool in terms of Loop adoption. Mr. Look replied: “Our definition of success is Tidepool Loop in an app store – initially in the US and international not long after that – and working with multiple partners’ CGMs and pumps, and demonstrating that Loop is a safe and effective way of delivering therapy with a vibrant ecosystem of devices. As for hurdles, the biggest one is getting all of the control and data protocols and business relationships done with the device companies in order to enable this.”

• Giving credit where credit is due, Mr. Look noted that Tidepool is standing on the shoulders of giants (the DIY community). He invited the audience into a time machine, showing pictures of seminal moments leading up to this moment including the first D-Data Exchange, Ben West’s breakthrough hacking of a Medtronic pump, his daughter’s first software-delivered insulin, the development of RileyLink and Loop, and Mark Wilson’s seminal 2016 D-Data talk. It was cool to realize that nearly every giant he alluded to was in the room, and all of this has taken place in a span of ~5 years!

4. FDA’s Bakul Patel, Pilot Reps (Verily, Fitbit, Tidepool) Provide Positive Update/Commentary on Pre-Cert; 2019 Test Phase of Pre-Cert V1.0; How to Overcome Inertia of “Old School” Regulation and Fear of Regulators?

Prior to an excellent panel on FDA’s digital health Pre-Cert program, Mr. Bakul Patel reiterated plans to release a working model (V1.0) of Pre-Cert by the end of 2018, leading to an important 2019 where the program will be piloted and iterated upon. Mr. Patel said this 2019 “test phase” will seek to find out if the firm-based Pre-Cert program can give the same or better assurance of safety and effectiveness as the current product-focused regulatory model. He didn’t say exactly how this would look, as we don’t imagine that FDA will have two groups working in parallel – one on firms and the other on products – to compare the strategies. In explaining the Pre-Cert program, Mr. Patel switched his metaphor this time around, going for an E-ZPass toll booth rather than TSA PreCheck – you still have to pay to get through a toll, but the transaction becomes “extremely smooth.” We appreciated another of Mr. Patel’s metaphors, which portrayed FDA before Pre-Cert as holding a finger in the dike, preventing the flood of digital health regulatory questions, but they hadn’t answered the question of what should be done when the flood surge becomes too strong. The underlying intent of this program, he said, is to have quality baked into every step of every organization (rather than maintained solely by one quality assurance manager) and to shift conversations from processes to outcomes in addition to processes. He added that he’d rather talk directly with the people who are building a product in an interactive way, rather than solely examining the paper submission documentation – to gain confidence in a product, it helps to have confidence in the manufacturer’s validation, processes, and organizational structure, on top of the product itself. Notably, FDA is starting small with software since “if you tackle the problem of the fastest moving object, you can scale it to other areas down the road.” We loved Mr. Patel’s response to a question about the consequences of not maintaining organizational excellence: “We haven’t built that yet. We’re focused on building this extremely easy path for people to follow… If I say follow the easy path and you intentionally don’t or have a malicious reason, then a Form 483 or paper letter doesn’t cut it for me…I refer back to patients. What should it be? 483? Something like a parking ticket for not parking in the right spot or driving too fast?” We also loved that when FitScript CEO Mr. Charlie O’Connell asked how he could get involved in Pre-Cert, Mr. Patel simply said “you’re already enrolled to help us…engage with us. Help us build it.”

• A major question during the session – initially posed by former Glooko CEO Mr. Rick Altinger, who was advised by a regulatory consultant to not bring Glooko into the Pre-Cert pilot – was how to overcome the inertia of old school mindsets around the role of regulators. Mr. Altinger said that following his regulatory consultant call, he spoke to the CEO of a diabetes device company who told him “No, stay away, that could bring all kinds of trouble.” He suggested that organizations with a lot to lose wouldn’t want to be involved in the program, while nonprofits (e.g., Tidepool) and those with less revenue tied to regulated products (e.g., Fitbit) would be more inclined to participate. Tidepool’s Howard Look pointed to J&J and Roche’s participation in the program, and “called BS” on the companies who said they wouldn’t participate because they are afraid of the FDA. “We all have a responsibility to push back on the old school way of regulating, it’s wrong for the product, it’s wrong for patients, and it’s wrong for the community.” Mr. Altinger’s point was more a reflection of what he had heard from others in the field, and he’s now in on Pre-Cert and excited to help push it forward, but we appreciate him bringing up the point that it could be hard to convince some of the more risk-averse players to participate. Mr. Patel said that it’s not as gloomy as Mr. Altinger thinks, as the day the nine pilot participants were announced, he got numerous calls from companies asking if they could get in after having initially said they would “wait and see.” Mr. Chris Bergstrom pointed out from the audience that it’s basic human psychology for most to not want to be first, but no one wants to be last.

• Mr. Patel seemed unphased by a letter from three US Senators who posed questions and expressed concerns about Pre-Cert, pointing out that most of the questions were pulled straight from the FDA’s documents. Mr. Look and Verily’s Ms. Afia Asamoah agreed that the questions were totally reasonable and the ones that FDA, Mr. Patel, and participants should be asking themselves when launching such a product. Ms. Asamoah took it a step further, stating that she views the letter as a positive sign that Congress is engaging with FDA and interested in keeping Pre-Cert credible and transparent, not so much that they disagree with its philosophy. The letter’s authors requested a response from FDA by Friday, November 9^th.

• We’ve compiled a number of notable quotes from the panel that we felt captured the enthusiastic and optimistic sentiment around the program from participants:

  • “As a consumer, I really value what FDA does. Regulatory oversight of medicine – having assurance that a medication is safe when I give it to my child at night – makes me feel really good. We at Fitbit believe in this process that will allow us to continue our speed of development, continue to drive discovery and health outcomes, and we can stand behind it with the weight of the US government. It’s valuable, allows us to work with other peers, Verily, other companies doing large health outcomes; we’re all doing it together and it’s being done right.” – Mr. Friedman

  • “Pre-Cert was a no-brainer for Verily to be a part of the conversation, to shape the way SaMD (software as a medical device) is regulated and get it onto the market quickly and efficiently. I worked at FDA for a while. I was there in 2010 when Bakul talked about the market and mobile technology. Our CEO Dr. Andy Conrad also felt deeply about FDA as an important regulator. [Former FDA Commissioner Dr. Robert] Califf is a Verily advisor. We see FDA as a critical institution for getting products on the market and making people healthier. The opportunity to participate was something we had to say yes to.” – Ms. Asamoah

  • “Back in 2013, we got great advice to meet with the FDA early and often. As a non-profit we had nothing to lose. We read the regulations and the guidance documents, some of which were written before the Internet existed, and said 'Hey, we think some of these guidance documents are stupid and don't make sense in a world of modern agile software development. We think we have a better way of building regulated medical software that we'd love to show you.' And they said 'Great!' and even 'Please share that with other startups!'” – Mr. Look

  • “When I first heard about Pre-Cert, I thought it was a fantastic idea. But it’ll be really, really hard to do. Culture change is incredibly difficult. The other point initially for me was how are you going to measure a culture?...One year later, I actually do believe you’ll come up with an excellence appraisal that’ll keep the bar high for safety and effective products but still allow for nimble development….We are invested, and I hope everyone else is, to be helpful to you because we want this to be successful.” – Ms. Asamoah

  • “Pre-Cert actually levels the playing field. It says follow these things that make sense to the regulator and also the product people and also the engineers and makes sense to both, and it allows those groups to talk well together. In Pre-Cert, you have Fitbit and Verily, but also a bunch of startups too. The diversity of perspective here has made it a lot of fun.” – Mr. Friedman

  • “Pre-Cert is actually going to make Tidepool’s processes better.” – Mr. Look

• We were delighted to hear Mr. Patel speak very intently on better incorporating the patient voice, opinion, and preference into regulation. In fact, he personally believes that the term “patient” labels people before they’ve actually been heard, and he wants to rethink FDA’s use of the term! He said he reads all patient comments that are submitted, and he touted ongoing initiatives at the Agency to “make patient input matter…how do you take preference and turn it into something that can be actionable?” At the same time, he noted that FDA’s “been screaming about patient voices in all policies for eons, but being at the Agency, we don’t see patients at public meetings. It’s not because they’re not interested, but resources are just not there. Verily’s Ms. Afia Asamoah believes FDA is very interested and engaged in hearing patient perspectives, but needs to figure out ways to make it easier for the public to engage. Well known T1 mom and advocate Ms. Stacey Simms echoed this sentiment, claiming that this is the first time in the 12 years since her son was diagnosed that she’s been in the room with someone from FDA. “My question is: How can we help?” she asked, “I have never thought to contact FDA. I have never thought you’d want to hear from us, that we could help. Is there a patient pre-submission meeting? Can I have you come to Charlotte, NC? I’ll have 100 people in the room.” Mr. Patel directed her to the Patient Engagement Advisory Committee meeting which will hold its first meeting on November 15^th in Bethesda. Notably, another well-known advocate, Ms. Moira McCarthy, said that the beautiful thing about our government is that they do actually listen, including FDA does listen to their people.

  • Mr. Friedman pointed out that patients should reach out to companies as well for two reasons. Firstly, saying “we’d love for you to do XXX” can impact a company’s direction, and secondly because FDA asks manufacturers if they’ve talked to patients/performed user studies/gotten user feedback. Ms. McCarthy advised companies who want Pre-Cert to go beyond pilot and become reality to “find your patient voices, that voice is the one that affects change.”

5. BCG’s Chris Bergstrom Keynote: Diabetes Care is Not Consumer Friendly Enough…Where Have We Been and How Do We Advance? Ethnography, Moving “Beyond the Widget,” and Asking for Help Embracing Agile Culture

BCG’s Mr. Chris Bergstrom kicked off DiabetesMine University with his take on the evolution of consumerization in diabetes care: Consumer-friendliness is not where it needs to be, but companies are investing in scale, partnerships, and the requisite “fail fast” mentality to move in the right direction. In the engaging keynote, he tapped into interesting stories from his past to explain the value of observing people “in the wild” to inform R&D agenda, discussed the commoditization overload in diabetes (and how that is good for patients from the perspectives of pricing and driving services and experiences “beyond the widget”), and some neat examples of BCG clients and other organizations (Allianz, John Hancock, Oscar Health, Qantas) and how they are differentiating themselves in insurance.

• Mr. Bergstrom took the audience “Back to the Future” through demonstrative anecdotes from his career and making the case for an ethnographic approach to market research. When he first arrived at Roche in 1999, colleagues showed him two meters: One was on the way out, and the other was about to be introduced onto the market. Apparently the device aesthetics didn’t make it clear enough as to which meter was new, as Mr. Bergstrom learned when he incorrectly assumed that the new meter was the one that was about to be retired! Fast forward to 2002, a landmark year in diabetes when companies began spending large sums of money on TV advertising. In Mr. Bergstrom’s view, this shift in marketing budget was an acknowledgement that the customer matters and they need to be served. In his experience at Roche, the hunch turned out to be true, as the direct-to-consumer marketing sent sales through the roof. And then sometime in the mid-2000s, “everything started getting painted in rainbow colors.” Color choices are great, Mr. Bergstrom said, but there are more important things to improve upon from an R&D perspective, and ethnography is a better way to figure out what those priorities should be than simply asking. Hearkening back to a famous Henry Ford quote – “If I had asked people what they wanted, they would have said faster horses” – Mr. Bergstrom displayed a now-amusing slide from a presentation he gave at Roche in 2001 that ranked blood drop size, test time, and strip integration as the top three patient priorities. Features like memory size, data management, calculations in memory, and transfer of memory results were much further down the list because people couldn’t possibly conceptualize how such improvements would better their lives. Ethnography, now more widespread but probably still underused, allows researchers to observe people in their natural habitats, look for friction and pain points, and use that information to develop product concepts that can then be tested and iterated on.

• Mr. Bergstrom sees a “commoditization overload” in diabetes since there are more than three major players in almost every drug, device, digital health, and services vertical, but this is generally good for consumers. A wealth of products that are not overwhelmingly differentiated leads to downward pricing pressures and importantly forces business owners to think differently about their value-add since “you don’t run commodity businesses for too long.” This is where “Moving Beyond the Widget” comes into play, he said, focusing more on the user experience than the product itself. “When products are easy to access and plentiful, the next lever is how to get the right products to the right people at the right time in a seamless way that is delightful and where problems that do arise are solved quickly (maybe even before the problem comes up).” Industry is keenly aware of this “last-mile problem” and committed to solving it, but Mr. Bergstrom noted that services, analytics, experience design – crucial aspects of going beyond the widget – are not in the typical pharmaceutical or medical device company skillset. The first step to progress, in his view, is asking for help. Diabetes may be a posterchild for taking bold steps to build agility into traditional contexts; look no further than Onduo (Sanofi/Verily joint venture), Novo Nordisk (Glooko and IBM partnerships), Lilly (Cambridge innovation center), Roche (mySugr acquisition), Tandem/Lilly /Roche / Dexcom (use/hire consultants from the DIY community), etc. “Integrator” companies are also taking on more and more facets of patient care, positioning themselves to offer coaching, address socioeconomic causes of suboptimal self-management, and ultimately place fees at risk. Through a consumer lens, greater commoditization has the benefit of increasing choice, but also contributing to overwhelm, me-too, and leaving companies with less resources for R&D.

  • As seen in the graph below, pharma has spent a very small portion of its revenue on technology in comparison with banks (but at least more than the construction industry). This may be one reason why banking can now be done almost entirely remotely from a phone (if not in an automated fashion); to pharma’s credit, while banking does bring concerns of privacy, it doesn’t face the same regulatory and safety hurdles. We’d like to know more about the time horizon captured in the graph, change of investment over time, the gross quantities of investment, investment by pharma disease area, definition of “technology,” and a number of other questions, but the figure is certainly intriguing. On the plus side, pharma is certainly investing a lot more in digital health in the past three years, a trend we expect to continue.

• BCG Digital Ventures basically brings a “startup in a box” to help organizations with entrenched cultures “accelerate what they probably couldn’t do on their own” in healthcare. Mr. Bergstrom provided a diverse list of clients making splashes in the insurance domain. Qantas, Australia’s largest airline, wanted to get into the health insurance market. In doing so, it launched its own health insurance company, Qantus Assure, that includes a neat wellbeing app that rewards customers with Qantas points – redeemable for Qantas (airline) points or a wide assortment of rewards – for physical activity as measured by a smartphone or tracker (similar in principle to the UnitedHealthcare Motion Program). Qantas Assure also maintains a dashboard of nationwide activity (at the time of writing, average step count for the day was ~4,000 in a group of 170,000+ participants). Less than 18 months after launch, Qantas Assure is reportedly the fourth largest and fastest growing health insurance company in Australia, pulling in 1oos of millions of dollars.”

• In doing so, it launched a neat wellbeing app that rewards customers with Qantas points – redeemable for Qantas (airline) points or a wide assortment of rewards – for physical activity as measured by a smartphone or tracker (similar in principle to the UnitedHealthcare Motion Program). Qantas also maintains a dashboard of nationwide activity (at the time of writing, average step count for the day was ~4,000 in a group of 170,000+ participants). Less than 18 months after launch, Qantas is reportedly the third largest health insurance company in Australia, pulling in 1oos of millions of dollars. (On a similar note, John Hancock recently announced that life insurance premiums will go up or down depending on a participants’ activity levels.) BCG also worked with Allianz, which wanted to make its way into the insurance market in China. Recognizing that people are most open to purchasing life insurance when they are starting a family, the company developed a home use fetal heart rate monitor (“Kaishi”) in hopes of building a relationship with a customer so that when they do come back for life insurance, they will choose Allianz. Lastly, Google-backed insurer Oscar Health is about to be cash flow positive and has innovative features such as free activity trackers, unlimited free telemedicine, and rapid claims adjudication (three weeks now, “near real-time” by next year)! These stories of aligning incentives and building delightful experiences for patients are always fascinating, and we’re glad to see so much movement on this front.

  • Other ventures Mr. Bergstrom highlighted included: PainScale (back pain app), Emilie Scientific (digital clinical trial recruiting, consenting, and on-boarding), Gaido (filing with FDA next year for software/wearable device combo that helps get people with cancer out of the hospital faster), and Oshi Health (an app to facilitate patient/provider communication for people with IBD).

6. Bigfoot’s Lane Desborough On Machine Learning: 20 assertions, value of virtual clinic/modeling approach to AID Development

Bigfoot’s Chief Engineer Lane Desborough shared 20 assertions related to machine learning and the company’s virtual clinic approach to AID algorithm development. They are listed in order below and reflect the company’s commitment to extensive modeling and simulation, rather than running a series of pre-pivotal clinical trials. We caught a glimpse of this at DTM 2017. To date, Bigfoot has done just one clinical trial prior to its planned 2019 pivotal. Mr. Desborough made a very strong case for the benefits of simulation: complete control of conditions, testing scenarios which would be dangerous with human, avoidance of recruitment bias, low cost, and speed. He also noted FDA is increasingly embracing simulation; simulation and clinical trials play complementary roles in the design and characterization of AID systems.

1. Bigfoot is a “Machine Learning“ (and “Data Science”) company. Lane’s favorite diagram explaining this (source):

2. Diabetes is a Data Disease

3. Living with diabetes would be easy if glucose never varied

4. ~42 things contribute to glucose variation – see Adam’s article here.

5. Experimenting with humans is dangerous, expensive, time consuming, and not very informative. To a chorus of laughter, Mr. Desborough showed a plot of messy glucose traces, documenting his son’s highly-varied blood glucose responses (n=304!) to the same juice box (16-grams of carbs).

6. Automation transfers variation from a place where it hurts to a place where it doesn’t hurt as much, in order to make a human’s job easier. In the case of AID, it’s about transferring variation from glucose to variation in insulin.

7. You can’t transfer variability if you don’t know the nature of the variability

8. The best way to build an automated insulin delivery system is to first build a simulation. “If you’re going to build a safe, effective system, you must build a simulator first.”

9. Simulators may be new to medical device development, but they are not new to other complex safety critical industries – e.g., chemical plants, aircraft development

10. Simulations contain models, which contain data, which contain variation

11. Diabetes Data is not Normally distributed – Mr. Desborough jokingly noted this fact is rediscovered every 5-10 years, citing publications from 1976, 1988, 2002, 2008, 2012. “If you are ever tackling a model of diabetes data, lognormal is the best assumption to start with.”

12. Models go through a lifecycle on their way to providing value.

13. vClinic is Bigfoot’s simulator, which we first learned about at DTM 2017. vClinic allows Bigfoot to run very large simulations very quickly. "Rather than run a clinical trial that takes months to understand a scenario such as inconsistent bolusing behavior, we can run a simulation that takes five seconds."

14. Simulation complements traditional methods of system characterization (source)

15. Simulations predict outcomes – see DTM 2017

16. Simulation is hard – “It’s taken us a tremendous investment in excellent people to build models and not make the common mistakes, such as overfitting and over parametrizing.” One of the big challenges, noted Mr. Desborough, is verification and validation – “how do you know the simulator is accurate and high fidelity?”

17. FDA is a strong supporter of Modeling and Simulation – “The disadvantages of human trials are complemented by modeling/simulation.” Mr. Desborough cited this guidance document as one example and several FDA conference presentations (here and here). “There will be an increasing role for digital evidence, which could accelerate approval of medical devices. Over time, the role of clinical trials will shift from exploration to confirmation that the models are correct.

18. Simulation enables simpler, safer designs – Mr. Desborough cited the graphic from T1D community member Ben Kamens (source), which provides a skiing analogy. It is really easy to add another feature to a device (green circle); it is quite hard to decide not to add features (blue square); hitting an MLB fastball is harder (black diamond); and removing features that have been shipped is next to impossible (double black diamond). Data science plays a key role in the minimization of complexity/prevention of the addition of features (which would be nearly impossible to remove once they are in). We also liked Mr. Desborough’s stereo analogy, which put the simplicity differently: stereos used to come with tons of audio knobs to tweak the sound; now, it’s just a volume dial on an iPhone. “It’s almost like the audio engineers didn’t know how to make good sound, so they gave it to the user to figure out. What data science allows us to do is to twiddle the knobs automatically, lock those buttons down, and remove them from the face of the device.”

19. Simulation is the only practical approach to developing a scalable automated insulin delivery system to reduce the burden of living with insulin-requiring diabetes. “We can predict outcomes over months or years, including the economic benefits, or our system, reduction of hospitalizations, etc.” Simulation has allowed Bigfoot to engage with investors and even payers in a data-based way. “You can do things to virtual humans that you cannot do to real humans.” Mr. Desborough also pointed out simulators can test a system in the actual intended user and real-world use conditions, rather than the highly engaged trial population that often comes to AID studies. “Plus, simulations are 4,000,000 times faster and less expensive than real-time clinical trials – 1 cent per simulated contact-day vs. $1,500 per contact-hour.”

20. #WeAreNotWaiting to use simulation to hasten the development of our systems

21. Bonus Assertion: Bigfoot is hiring for the most mission-driven (and data-driven!) team in diabetes

-- Brian Levine, Adam Brown, Sarah Kolk, and Kelly Close