DiabetesMine Innovation Days 2020

November 5-6, 2020; Virtual; Day #1 Highlights – Draft

Executive Highlights

  • DiabetesMine’s Innovation Days 2020 got underway Thursday morning Pacific time. As usual, DiabetesMine’s Amy Tenderich and her powerhouse team have put together an excellent program and powerhouse group of moderators, presenters, and thinkers on diabetes technology.

  • Thursday’s sessions had a pointed focus on digital health and software. In the morning’s panel, we heard from several perspectives, from payers, patients, and entrepreneurs, about the role of technology and coaching in diabetes care. We also heard from co-founder and executive director of the Digital Medicine Society, Jen Goldsack, on some of the concerns and goals she has for digital health and medical technologies.

  • In the afternoon, we were treated to survey results and a panel aimed at highlighting the voices of Black, indigenous, and people of color (BIPOC) in diabetes (check out a powerful 5-minute video put together by DiabetesMine here). In a convenience sample of 207 BIPOC living with diabetes, more than half (51%) of respondents reported that they do not feel represented when they see commercials or websites for diabetes devices and treatments. Notably, rates of diabetes technology use were high among respondents, much higher than is seen on average in the field, suggesting that they are more educated on diabetes than is usual. Sepcifically, 80% of survey respondents reported using CGM while 71% said they used insulin pump therapy. Following the panel, Dr. Brynn Marks highlighted the fact that the Control-IQ adult and pivotal trials included no Black participants and suggested some steps to help make clinical trials more accessible.

Top Four Highlights

1. Holistic Diabetes Care through Technology: Leaders from One Drop, Vida Health, January.ai, and Healthfirst Talk CGM, Telemedicine, Coaching, Artificial Intelligence, and More

During the first panel of the day, Dr. Susan Beane (Healthfirst), Ms. Noosheen Hashemi (January.ai), Ms. Rachel Sánchez-Madhur (One Drop), and Ms. Susan Tilenius (Vida Health) embarked on a wide-ranging discussion on technology and diabetes care. DiabetesMine founder Amy Tenderich moderated the panel and represented the payer perspective as the Executive Medical Director for  Healthfirst, a New York HMO that has contracts with Cecelia Health and Livongo for diabetes management. Ms. Sanchez-Madhur represented both the digital health startup perspective, as SVP of Product at One Drop, and also the patient perspective, as someone with diabetes. Finally, Ms. Hashemi and Ms. Telinius brought their perspectives as leaders in digital technology and founders and CEOs at their companies, January.ai and Vida Health, respectively. The four female panelists offered thoughtful insights into the importance of integrating considerations for personalization, patient empowerment, and accessibility.

  • Not surprisingly, all three startup panelists highlighted the value for AI and data analytics to provide more personalized and meaningful insights for people with diabetes.

    • For One Drop, Ms. Sanchez-Madhur emphasized an approach, using data analytics, that predicts users’ short-term (e.g., hourly glucose predictions) and long-term health status (e.g., A1c predictions months out). Together, these predictions could be used to give users the single most important and actionable step they could take each day to improve their management.

    • At Vida, Ms. Tilenius outlined a similar approach: using their company’s models to predict how effective Vida’s intervention might be in specific users. Then, using those predictions, Vida can explore the various barriers preventing users from successfully lowering A1c all to the end of working out how to individualize advice on how to best prevent the barriers.

    • Finally, January.ai’s Ms. Hashemi discussed data analytics from a different angle, noting that the sheer amount of data that can be generated and recorded (e.g., CGM, BGM, food logging) is impossible for humans to meaningfully sift through and understand. Instead, armed with the data, algorithms and machines could do the work for us in personalizing recommendations and care for individuals.

    • “One Drop is developing algorithms that both predict long-term and also short-term health. We can predict blood glucose over the next 24 hours. We can also predict where you’ll be and what your overall health state will be over months to a year. We’re trying to build turn-by-turn navigation for life based on your data and your health profile and assessing each day, what the most important thing is for your health that day. Let’s say you’re on low-carb diet, but also trying to reduce saturated fat and sodium. People just don’t understand which way to go or what to focus on. In taking a holistic approach – how you’re doing with blood pressure management, looking at diabetes management – we’re able to tell you what the most important this is for you today when you open the app. Maybe that’s low sodium, not low-carb. Maybe it’s more exercise. Our algorithms are able to give this feedback to you: what’s most important and what’s going to drive change.”One Drop’s Ms. Sanchez-Madhur

    • “Simply put we use AI to help people manage their blood sugar. We model their glycemic response to 16 million foods in our food atlas so that we can answer the question, ‘What should I eat?’ . We are optimizing for low glucose foods, but also for fiber. It’s about helping people understand their body and gain agency over their own body and become the world expert on their own body. Today, a doctor might prescribe a statin to lower cholesterol but that might raise A1c, so we basically support them in doing a bunch of experiments to determine what impacts what and then support them in doing the things that work for them. What we’re talking about is personalization and making it work for the person. We take into account how much water did they drink, how much did they sleep, etc. We want to learn their baseline and then help them figure out what needs to be adjusted … People can’t see inside your body and know what oatmeal will do to them. But we need people for all the feelings, and to have empathy for us on the journey that we’re on. And machines will not be replacing humans for any of that any time soon. But for knowing about what your food does, you can have a machine to do that.” – January.ai’s Ms. Hashemi

    • “One of the things we do is use machine learnings to watch and monitor what’s going on. With data from devices and conversations with providers, we have predictive models on whether you can lower A1c. With our program, we see A1c reductions of >1% over 3-6 months and we want to make sure people can do that. Social determinants, nutrition, and access to care are often the biggest drivers of success, so we want to watch those. We promote low-carb, Mediterranean diets and we have food tracking in the app, as well. But we use data to inform a holistic approach: is it nutrition, social determinants – what’s driving the barriers?” – Vida’s Ms. Tilenius

  • There was an emphasis throughout the panel on the characteristics of people that are successful with digital diabetes technologies and trying to expand that population. Healthfirst’s Dr. Susan Beane brought up the importance of creating platforms that are accessible and easy to use regardless of the barriers in the users’ life. During the panel, she noted that because a high percentage of Healthfirst's 1.5 million members may at some point experience a barrier to accessing care, Healthfirst seeks partners who offer their programming in many languages and that prioritize a range of culturally sensitive approaches. For example, “a population we’re working with now are people who immigrated from Haiti and speak Creole and have never heard of the diabetes care options we’re providing.” Additionally, Dr. Beane noted that these populations not only haven’t heard of the many care options, but might also be discouraged or intimidated, highlighting the importance for people with diabetes to have someone supporting and helping them. To this end, Ms. Sanchez-Madhur and Ms. Tilenius discussed some of their efforts to improve accessibility of their digital offerings.

    • “We do have a full Spanish offering. We do have offerings down to a fifth-grade reading level, so that we can do Medicare, Medicaid, and other groups. We are trying to figure out how to map [our materials] to the different cultures and we’ll continue adding languages as we need. We look at social determinants, loneliness, food deserts, and other factors. We have case managers who will help find access to food and things they need in their local area.” – Vida’s Ms. Tilenius

    • “We offer everything, so text, audio, video inside the app. Typically, we encourage a weekly coaching or therapy session, especially for therapy. People are controlling their own destiny in the app. They have 24/7 access to a coach, they can set up video session – they have the freedom. They get a lot of info and lessons and little nudges and insights inside the app, as well. It’s really up to them. Some are heavy trackers and like all the insights and the data. Others love having more consults with coaches. It’s really up to them and what works for them. We also have social groups to support each other inside the app, suggesting tips to each other. It’s about what works and motivates them.” – Vida’s Ms. Tilenius

    • “We are obviously available in every language we can be, right now 10 languages and expanding to more. We’re digital, so as long as you have a smartphone you can start using it. One Drop is incredibly accessible, available worldwide, with food libraries from all over the world. When we get to the question of ‘Who’s going to be successful?’ I think the first questions is: ‘Are they ready?’ Readiness to change is a really important factor. Those who are ready and motivated are going to be successful. Motivations can be different. It can be family members, recent interactions they’ve had with the healthcare system, suddenly they’re realizing something could go wrong, so they’re going to actively manage their disease. For myself, the biggest change for me was realizing that I could take control of my diabetes on my own and didn’t have to be a passive recipient of doctors’ advice. I could look up new drugs or devices on my own. Before, I didn’t feel as empowered to do so. It’s about being ready and being there and having the right motivation to keep going.” – One Drop’s Ms. Sanchez-Mudhar

  • During one of her answers, Ms. Tilenius shared a worrying statistic about mental health and diabetes during COVID-19. Typically, before COVID-19, Vida Health saw about “17%-20%” of its new users with diabetes also have co-occurring anxiety and/or depression. Since the COVID-19 pandemic, that number has more than doubled (for Vida’s population) to “45%-60%” with many external factors such as concerns about health, economic uncertainty, and the 2020 elections weighing on people’s minds. On last week’s Teladoc 3Q20 call, we heard about a skyrocket in demand for mental health services among the broader population (i.e., people with or without diabetes): in Teladoc’s B2B channel, mental health service utilization grew by “over 500%” YOY.

2. DiabetesMine BIPOC & Diabetes Technology Survey: More than Half Don’t Feel Represented in Diabetes Marketing, High Rates of Tech Utilization; Panel of Advocates Talk Representation, Misdiagnosis, and Health Care Experiences

DiabetesMine dedicated significant time and attention on Thursday to the voices of traditionally under-represented populations, specifically Black, indigenous, and people of color (BIPOC) in diabetes. The afternoon sessions kicked off with quantitative and qualitative results from the Diabetes Mine BIPOC & Diabetes Technology survey (check out a powerful five-minute video put together by DiabetesMine here). The survey took place in September 2020 and collected responses from 207 BIPOC individuals, over the age of 15, who either have diabetes or cares for someone with diabetes. As a convenience sample, the speakers stressed that the survey responses are not statistically representative, but still, the responses provide powerful insights into the experiences of BIPOC living with diabetes. Of respondents, 91% reported using insulin with 9% of caregivers for people with diabetes using insulin. Additionally, from a gender perspective, disproportionately more respondents – 74% – were female.

  • On an encouraging note, rates of diabetes technology use were high among respondents. 80% of survey respondents reported using CGM while 71% said they used insulin pump therapy. Also encouraging, only 7% of respondents said they were unaware or aware but not using available diabetes technologies. However, people willing to participate in an online diabetes survey are also almost certainly individuals with greater access to diabetes education and not representative of the larger population of people with diabetes. When asked about their sources for information on diabetes technology, 45% of survey respondents cited their endocrinologist, while 14% credited the diabetes online community. Additionally, though CGM and pump use rates were high among the survey population, 44% of respondents “had issues” that interrupted their use of diabetes technology including: (i) insurance issues/cost of technology; (ii) device malfunctions; and (iii) lack of communication between medical suppliers and providers.

  • Respondents were also asked about their experiences with HCPs, with 65% reporting they felt respected by their providers and 60% saying they felt listened to. However, only 37% reported reeling comfortable asking questions, 18% felt talked down to, 14% felt stereotypes are at play in the care they receive, and 13% feel disrespected. While these results indicate the majority of respondents have positive experiences with HCPs, the percentages who are uncomfortable asking questions or feel disrespected are still worryingly high. Additionally, when respondents were asked about their worst experiences with diabetes care, respondents shared about: (i) lack of empathy among HCPs; (ii) HCPs who were not well-informed culturally; (iii) misdiagnoses (e.g., assuming Black people were type 2s); (iv) insurance-related barriers; and (v) challenges upgrading to new technology.

  • Survey respondents also provided qualitative answers to questions about perceived differences in experiences of BIPOC and White people with diabetes. Notable themes included a lack of understanding from doctors about patient experiences; disparities in education, information, and access to mental health resources; a lack of clarity on different diabetes diagnoses; and an overwhelming lack of BIPOC representation in diabetes advertising. In fact, more than half (51%) of respondents reported that they do not feel represented when they see commercials or websites for diabetes devices and treatments. Additionally, while 22% of respondents said diabetes companies had reached out to them to be part of a team, committee, or study, 27% of those respondents “felt like it was more for show/making a quota” than out of genuine care for their opinions.

  • Following presentation of the survey results, diabetes blogger Ariel Lawrence moderated a panel discussion between BIPOC diabetes advocates including actor Kyle Banks (founder KyleR Cares Foundation); diabetes and health author Mila Clarke Buckley; Joyce Manalo, a Senior Community Health Worker at Parkland Health and Hospital System; and DCES Quisha Umemba. Themes included the lack of representation of BIPOC in diabetes related media, as well as the need for technology developers to truly listen to BIPOC patient voices, not just “tokenize” them. Ms. Manalo discussed her experience as a community health worker acting as a liaison between patients and providers. She highlighted the need for improved provider communication and an emphasis on “approaching conversations from a place of learning.” Ms. Manalo also encouraged providers to take a step back and examine their own biases before providing care to make sure biases don’t influence the quality of care provided. Mr. Banks also emphasized the importance of patient education and patients being able to advocate for themselves. Ms. Umemba echoed this sentiment, saying patients need to be empowered to “be the CEO of [their] health and [their] life.”

    • During the panel, Ms. Buckley shared her experience with misdiagnosis to demonstrate the importance of providers listening to patients and recognizing their own biases. Unfortunately, Ms. Buckley had to see three providers before she was able to get a referral to an endocrinologist. Ms. Buckley’s first two providers were convinced she had type 2 diabetes, despite the fact that she was not improving with treatment and behavioral changes. Following referral, Ms. Buckley was instead diagnosed with LADA by her endocrinologist and is now receiving the therapy she needs. Ms. Buckley’s experience was emblematic of stories shared both in the DiabetesMine BIPOC survey. Unfortunately, Ms. Buckley’s experience provides an example of what can go wrong when biases and stereotypes influence care.

    • As the panel discussion drew to a close, Ms. Umemba outlined her three guidelines for improving care for BIPOC patients: (i) “refer everyone period”; (ii) full disclosure on all diabetes treatments and therapies available; and (iii) frequent and often follow-up. Specifically, Ms. Umemba recommended providers refer patients to nutritionists, dieticians, and community education programs regardless of whether or not the providers believe patients will utilize the resources because patients will not have the opportunity to use them if they are not first referred. Ms. Umemba also related this to Ms. Buckley’s misdiagnosis story in which it wasn’t until she was referred to an endocrinologist that Ms. Buckley was correctly diagnosed. Ms. Umemba also advocated for providers to be fully transparent and outline all available diabetes therapies and technologies so patients can participate in the decision-making process which has been proven to improve outcomes. Finally, Ms. Umemba stated that care should not be “follow up as needed, [but] follow up is needed” because without frequent follow-up providers are unable to track patient outcomes and provide early interventions.

3. A Need for More Inclusive Diabetes Technology Studies: Recent AID Trials Report >80% White Participants, 0% Black Participants; Children’s National Hospital Running AID Study in Minority Groups

Following the BIPOC panel, pediatric endocrinologist Dr. Brynn Marks (Children’s National Hospital) provided an overview of racial diversity (or lack thereof) in landmark diabetes technology studies. In the landmark iDCL3 trial – the adult pivotal for Tandem/Dexcom’s Control-IQ – 89% of participants were White. Additionally, over half of participants in that trial had households over $100,000 (62%) and nearly all were privately insured (93%). That study was presented at ADA 2019 and published in NEJM. In the DCLP5 trial – the pediatric pivotal for Control-IQ – 81% of participants were White, 8% were Hispanic or Latino, 2% were Asian, and 0% were Black. That study was presented at ATTD 2020 and also published in NEJM. In the pivotal for Medtronic’s MiniMed 670G (presented at ADA 2016 and published in DT&T), there were no details on racial/ethnic makeup of the sample. Regardless, the lack of representation in these trials makes it difficult to accurately generalize results to minority groups – Dr. Marks recalled a study presented at ISPAD last month that found lower rates of CGM initiation and higher rates of CGM discontinuation in Black and Hispanic children, relative to white children.

  • On a more encouraging note, Dr. Marks also shared that Children’s National Hospital is currently working on a study of Control-IQ, targeting underserved populations. The study will enroll ~30 type 1s, ages 6-21, focusing on non-Hispanic Black children, children publicly insured, and those with A1cs ≥10%. Additionally, carb counting is an exclusion criterion for the study, meaning only those on sliding scale or fixed dose regimens at baseline will be included. The six-month will use Time in Range as the primary outcome, while also looking at adverse events, treatment satisfaction, and quality of life. Importantly, participants and their caregivers will also be interviewed to assess how the technology affects their outlook and diabetes experience.

  • Dr. Marks also presented some advice for making research more accessible. Firstly, she called for providing more compensation for studies, as most study compensation amounts are not enough to make up for lost productivity and wages. To this end, Dr. Marks also suggested making remote and night-time study participation more common. Further, Dr. Marks discussed the issue of mistrust of the medical community, particularly by Black people. With the disclaimer that she “doesn’t have the answers” to this issue, Dr. Marks did note that, in her experience, children respond better to care teams that look like them and recruiting more diverse providers and study investigators might lead to more diverse participant recruitment, as well. 

4. Digital Medicine Society Boasts 1,600+ Members One Year After Launch; Society’s Jen Goldsack Discusses Balancing the Power and Risks of Digital Health and Medical Technologies

During Day 1’s opening presentation, Jen Goldsack, the co-founder and executive director of the Digital Medicine Society, discussed the importance of balancing the power and risks of digital health and medical technology. This issue has becoming increasingly important as technology’s role in healthcare has grown with healthcare’s monumental shift toward remote patient monitoring and telehealth. Just in recent months, three of the biggest five companies in the world (Amazon, Alphabet, and Apple) announced new health monitoring products, (Amazon Halo, Alphabet’s pending acquisition of Fitbit, and Apple Watch 6). Ms. Goldsack launched the Digital Medicine Society (DiMe) in May 2019, because she and others believed in the power of technology but wanted to ensure that it is a tool for good and not a weapon: “DiMe exists to advance the safe, effective, equitable, and ethical use of digital products to optimize human health.” The Digital Medicine Society “sits at the intersection of two communities: healthcare and technology” and includes over 1,600 members from 43 countries with members from regulatory and government bodies (e.g., FDA, NIH, Veteran Affairs), industry (e.g., Dexcom, BD, Pfizer, Novartis, Roche, Biogen, AstraZeneca), associations (e.g., Digital Therapeutics Alliance, Biohacking Village); academia (Yale, UC San Diego, Scripps, Duke, Northumbria University (in Newcastle upon Tyne), UNC, and patient groups (e.g., Savvy Coop). Patients are invited to join the society as well, and patient perspectives are central to their organization’s work. “We take a three-tiered approach to patient involvement: (i) our membership is open to patients, and they have equal voices and equal membership; (ii) in our research, we start with the patient voice and focus a portion of our research directly on patient engagement; and (iii) we have formal partnerships with patient organizations like Savvy Coop.” More broadly, the organization focuses on research, communication and education, and community building with over 26 publications since founding last year, eight public comments submitted to the federal government, and 47 sponsors collecting digital endpoints (~20% of which are related to diabetes) – see the full list of digital endpoints here.

  • During her presentation, Ms. Goldsack used an interesting phrase, “The Internet of Bodies”, to describe new medical technologies. The phrase is, of course, an extension of the Internet of Things, that specifically includes the network of devices that are ingested, implanted, or connected to the body. In diabetes, the Inter of Bodies would include technology like CGM and insulin pumps, but also things like fitness trackers and food loggers.

    • On the Internet of Bodies, Ms. Goldsack expressed concerns about security: “The evidence base for the safety and effectiveness of these new products has not kept pace with their development. Given the great divide between the promised benefits of digital medicine and its potential risks, we need to know – not just believe – that the tools we use are trustworthy.” Ms. Goldsack also quoted Forbes journalist Bernard Marr, “The security challenges faced by Internet of Bodies tech are similar to what plagues Internet of Things generally, but there can be life and death consequences when IoB devices are involved.” To this end, we know AID might be one of the first places where cybersecurity in medical devices is really tested – see our coverage of an FDA meeting on the topic from 2019 for more.

    • Ms. Goldsack also addressed the question of privacy: who has access to your data and for what purpose, noting that “If you are not paying for it, you’re not the customer; you’re the product being sold.” We suggest checking out a World Economic Forum article by Xiao Lui, “The Internet of Bodies is here. This is how it could change our lives,” and The Wall Street Journal article “The ‘Internet of Bodies’ is Here. Are the Courts and Regulators Ready?” by Andrea M. Matwyshyn for more information on these issues. It’s a complicated issue and we are by no means experts, but we found Ms. Goldsack’s presentation a good reminder that these are issues about which we should be thinking. If you’re interested in these questions and wanted to get involved in discussions and research, the Digital Medicine Society is always accepting members!

  • Ms. Goldsack also brought up disparities and racial bias present in medical technologies. Much in recent months and years has explored this issue. For example, a STAT article in 2019 discussed disparities in the accuracy of digital sensor technologies, looking at Fitbit and other wearables that may not accurately track heart rates in people of color. Similarly, racial bias has also been found in healthcare algorithms. A 2019 Scientific American article discussed racial bias found in a healthcare risk algorithm, which resulted in black patients losing out on critical care. Importantly, this model did not explicitly consider race as a variable – instead, as a machine learning algorithm, the model became biased simply through the data it was being trained on.


--by Hanna Gutow, Katie Mahoney, Albert Cai, and Kelly Close