David Klonoff: 0:15

Hello, this is Diabetes Technology Report. I'm David Klonoff. I'm an endocrinologist at Sutter Health and UCSF. I'm here today with my co-host, David Kerr, who has two very special guests. I'm going to introduce David and he will begin the interview.

David Kerr: 0:33

Thanks, david, and hello to everyone again. David Kerr, speaking once more from Santa Barbara, california, and we have two superstars of diabetes research and technology on the podcast today. We're very, very excited. We're going to hear about a new innovation that's planned through the Diabetes Technology Society. So welcome Shahid, welcome Juan. It's lovely to see you today and to hear from you and just to really set the scene. I'd like both of you just to describe what's your day job, really set the scene. I'd like both of you just to describe what's your day job. But, more importantly, how did you end up being interested in diabetes in general and diabetes technology specifically?

Shahid Shah: 1:17

Shahid, do you want to go first? Sure, yeah, I'm a software engineer by training and I've these days mostly doing entrepreneurship, in which I've built and sold a few electronic health record companies digital technology but a lot of my work has been in very specific types of medical devices, you know, like insulin pumps or respiration systems, and what we deal with a lot is both in my own family as well as a lot of the patients that we've helped take care of with respect to our medical devices is always related in some way to diabetes and other chronic conditions. And so just a few years ago, dr Plonoff and I met at an FDA event, of all places, and he started talking a lot about what the Diabetes Technology Society was doing, and I was always interested in diabetes just from a healthcare perspective, but the fact that there was an actual society which just focused on technology around diabetes was really what got me. He had me at hello when he said technology and diabetes in the same sentence, and that's how we got together.

David Kerr: 2:17

He's very persuasive indeed. Juan, is your story similar or how did you? Is it different?

Juan Espinoza: 2:23

There are definitely some similarities. So, uh, my day so I'm a general pediatrician by training an informaticist, my day job is I'm the chief research informatics officer. Uh, and in robert h larry children's hospital, chicago, um. And I got interested in diabetes, uh, from the perspective of it being a condition that was very much probably at the forefront of the incorporation of technology and data for disease management, especially if you think about 10 years ago. Like no other disease used devices and data to manage it.

Juan Espinoza: 3:03

Diabetes is this highly quantified disease and has been for a long time, and so for me it came along in this professional side of like oh, this is an interesting problem, I like data and technology. Here's a disease with a bunch of data and technology. Let me focus on it. I was involved early on in efforts around CGME, hr integration and sort of thinking through innovative ways to develop uh, uh, develop care models for diabetes, um, working closely with the endocrinology uh division at children's hospital Los Angeles, where it was at the time, and we published a paper, uh about our work and building the first um CGM EHR integration. And one day I get an email from a guy named David Klonoff said like oh, I saw your paper, that's really interesting. We should talk and then you know, four or five years later, of fruitful collaboration. We're here now working on this project.

David Kerr: 3:54

Excellent. So this project, I mean, I take it. You mean the diabetes technology meeting just the other week in Burlingame. There was a huge amount of excitement about something that was new and being launched, and you guys are starting this. So, to avoid any delay, what is this project all about? What's this new project? Shahid, do you want to start off?

Shahid Shah: 4:17

Yeah, I'll just tell you from a quick tech perspective, but really the focus is on clinicians, so we'll let Juan pick it up from there. But really the focus is on clinicians, so we'll let Juan pick it up from there. The Diabetes Research Hub is a core technical platform where guys like me, as software engineers, are participating with real clinicians. Like all of you on this call to really focus our efforts on technology designed specifically for researchers who are driving towards cures and treatments. A lot of other technology that you see out there like, for example, when we create digital health systems for, like electronic health record solutions or patient portals many of those are focused on the patients themselves or the doctors or the hospitals and the health systems.

Shahid Shah: 4:58

But there's really not much out there, especially in diabetes, which says I'm building something for and by the researchers themselves, whose job it is to have real evidence-driven research, evidence-driven experimentation, evidence-driven data collection and so that evidence-driven part of it which just drives a lot of what clinical trials need to be done, whether they're randomized or not, or they're dealing with real-world evidence it's really hard for researchers to find done. Whether they're randomized or not, or they're dealing with real world evidence, it's really hard for researchers to find a place where they can call their own and do real work, and so that's what interested me is that most of my prior career has been focused on healthcare delivery professionals or patients and their caregivers and families, as opposed to researchers. But when I ran into, you know Dr Klonhoff and Dr Espinoza and these guys, they're like doing serious, serious data work specifically on the clinical side, and that's what excited me. So, juan, your view of this is even more important than mine. I think they complement each other.

Juan Espinoza: 5:59

So you know, I'll answer that question, david, by telling a little bit of a story, which is in 1999, the FDA approved the first ever continuous glucose monitor in the United States no-transcript. So it probably took until about the mid-2010s to really achieve both saturation, both from the perspective of the professional societies, the way that clinicians felt about these technologies, and really incorporating CGMs even into our clinical practice guidelines, which is normal for a new technology. Um and then, but that popularity, as as CGMs became more and more used. Not only were they being used in research about CGMs for clinical care and improving the clinical care of patients, but it started being used for other types of research.

Juan Espinoza: 7:14

Uh is thinking of the CGM as not just a tool in the clinical management of persons with diabetes, but CGM as a glucose sensor, as a new type of data source that is available to inform any kind of scientific query in diabetes and outside of diabetes. And so, as the devices became more available in clinical use, they also started becoming used more in research, and not just in research about validating them, analyze, store and share this essentially new type of data. Right, like we know how to deal with survey data, we know how to deal with data from other types of medical devices, we know how to deal with EHR exports, but this is essentially a new type of data that has new types of requirements. And so we feel really passionately and over roughly a decade or so of research in the community, that there are gaps in our knowledge right of how best to manipulate this data, and there have been, absolutely.

Juan Espinoza: 8:41

There have been publications where people have proposed new types of metrics, new types of methods for statistical analysis, but that information isn't necessarily well disseminated or understood across the entire research community, and I think it's a real barrier, right, it's a barrier for early stage investigators, it's a barrier for, maybe, clinical researchers who are less technical, and so it keeps people away.

Juan Espinoza: 9:09

It keeps people away from using CGMs, it keeps people away from using this data, and we think that this is really powerful, that this has the potential to change other diseases and other research modalities, just like it did in the care of patients with diabetes, and so the problem that we're most interested in solving is how do we demystify the technical piece of this, how do we make it easier for researchers to use CGM data? How do we aggregate the collective knowledge of a community of researchers over the last 20 years into evidence-based recommendations around how we analyze the data, how we clean the data, how do you deal with missingness? What type of imputation should you use? Which metrics are the best metrics to use for different kinds of applications? And I think by building both the tools the technology, which Shahid was talking about, as well as the community of researchers who can come around those tools and how to best use them to support research is really critical. I think fundamentally is what the Diabetes Research Hub is all about. The technology is critical, but it doesn't do anything without the community and the shared knowledge that then makes it move forward.

Shahid Shah: 10:28

Yeah, and one of the things we're doing right now, in a very similar way as if we would do if we're doing an entrepreneurial startup, is we're trying to build a very basic amount of technology just to start, because we can, and the technology is already ubiquitous and available today.

Shahid Shah: 10:39

But the thing that Juan and Dr Klonoff and others at DTS are teaching us is how do we actually establish a set of surveys where we go out and ask researchers what do they need, rather than just creating something and telling them what they should want?

Shahid Shah: 10:54

Right, and so you know, there is this idea that, especially, you know, arrogant entrepreneurs like myself we're like we're going to build it and they will come. And while that might be useful, when you already know everything in the market, in the case of research, the whole point is no one knows what the answers are, and that's why research is necessary, and so that's a huge area and that's a huge call to action for everybody. Listening to this podcast is if you are even tangentially related to the research field in diabetes, please reach out to us, because we actually have real formal surveys, actual interviews, or we're actually sitting down with the researchers asking them what are you doing today with CGMs and how are CGMs working well for you or not working well for you? And so, juan, talk about that important really part, about you know getting to the meat of what researchers are working on first, before building all the technology.

Juan Espinoza: 11:46

Yeah, no, that's absolutely right, Shahid. I think, as you said, we don't have the hubris yet to decide that we know best, and so we're really starting from a ground up approach where you know we're not editorializing, you know we're trying to find every possible metric, every possible analytic, every possible imputation method, every possible application, aggregate it, collect the information and build tools that help address those use cases and so. But to do that, you know we are actively trying to engage members of the research community through surveys, through interviews, through focus groups, and so there's opportunities to. So we do have a website already set up for the Diabetes Research Hub. I'm sure we can include the link in your show notes and on the website and so individuals can sign up to participate, to share their feedback and to tell us more.

Juan Espinoza: 12:48

And, honestly, if folks want to reach out to us in a more unstructured way, just quite literally send an email like hey, I'd love to talk to you about this and tell you about how we're doing things we want to know, because there's a range of researchers, there are people who are incredibly sophisticated and they're like oh yes, we're using these complex technical tools to extract the data. It all goes into this virtualized machine where we've built a bunch of code that structures are normalized and applies different metrics, and that's fantastic. That's really important research. Fantastic, that's really important research. And there's also the folks who are much closer to the patients, much closer to the clinical part, who have really critical clinical questions that they think data from the CGM is going to help them answer. But they don't have those technical resources. They don't have a PhD data scientist on their research team, they don't have an engineer to manage all of this data for them, and so you need to be able to solve both of those problems.

David Kerr: 13:44

Let's follow the dream here. So if this is successful for people with diabetes, for health, equity, for access.

Juan Espinoza: 14:00

Do you think it's going to be game changing? Yeah, that's a really good question. I'll tell you, my aspirational answer to that is yes, the way that it'll be game changing? Right, because we're working at the level of supporting research. So I think it'll be game changing in a couple of ways more researchers of different backgrounds and different level of maturity of, of, sorry, of technical maturity, meaning that they do or do not have data science expertise or data handling expertise. Um, this will be something that will be useful for folks who are, whether they're from coming to diabetes research from a physician background, from a psychologist background, from a social work background, right Exp expanding the pool of people who can do meaningful research using CGM data. So that's one.

Juan Espinoza: 14:49

There's increasing access and reducing the barriers to entry for the research community. The second is that, and a lot of the focus on CGMs historically and the research has been on, what are those key clinical metrics, the components of the AGP, the things that the manufacturers put in their portals, the things that we build clinical guidelines on. All that is really really important and it moves slowly for good reasons, right the need for the robust level of data, the need for appropriate regulatory approvals, for things for consensus development. That said, in the research world there's the opportunity to move a lot faster to help each other answer different kinds of questions and new questions that maybe are more relevant to specific subsets of patients and populations, or in specific aspects of conditions that may otherwise not rise to the same prioritization structure as some of these.

Juan Espinoza: 15:49

Bigger like this is how we manage all people with diabetes, but we can think about more specifically within a research context, of how to generate the information that might help us change the way that we make clinical decisions about the disease.

David Klonoff: 16:05

Juan and Shahid. To what extent do you think that the Diabetes Research Hub is intended to help an individual researcher understand their own data better, and to what extent do you think it's important or desirable for them to upload their own data and share data with other researchers?

Shahid Shah: 16:26

Yeah.

Shahid Shah: 16:26

So the way to think about that is that we have already built out enough code so that you could do both right, if you don't know how you want that data to be used, but you have some data to share like if Juan is better at it than you know you are you can upload your data and Juan can do the analysis.

Shahid Shah: 16:45

But at the same time, as soon as you upload the data, many of the things that Juan has already talked about we already do as part of the metrics management. So as soon as you upload data today, we'll do a whole bunch of standard calculations, standard algorithms and things like that, but then open it and make it ready for everyone else to use. So the good news is, as of today, it's already in good shape to do both of what you're saying. But the more we get like right now we're in a network problem, which is, you know, we've got about 10 or 15 studies that we've got the data in and we're trying to load this in. We'd like to be at hundreds of studies, so that those who are good at collecting data can get data, but those that are good at researching can actually research, and so we get the best of both worlds.

Juan Espinoza: 17:27

Yeah, and I think what I would add to that is you know, from a philosophy perspective, our goal is to advance the field, and I think we advance the field first and foremost by meeting researchers where they are and helping to solve practical problems that they face today. They are in helping to solve practical problems that they face today, and so a lot of our focus has been on creating the software tools that researchers can use locally on their own machine, on their own data, without necessarily sharing it back to us, because, you know, there might be different institutional reasons why they can or cannot, and that's okay, we are here. There are lots of good reasons why there are rules around that. Some of them are federal, some of them are institutional, and we're not asking we should comply with those rules, just like every other research repository complies with those rules.

Juan Espinoza: 18:10

So what we're saying is let us help you solve your problems. By the way, if you solve your problems using the tools that we provided you, you can then just very easily upload your data to the repository that we're building. You can then just very easily upload your data to the repository that we're building and that potentially solves two different problems for the researcher. Right, it solves. You needed better tools to analyze your data. Two, maybe this is an NIH-funded research study where you're required to deposit your data into a public repository. We are a public repository where you can deposit your data, so we're trying to solve two different problems and now that you've deposited your data by the're trying to solve two different problems and now that you've deposited your data, by the way.

Juan Espinoza: 18:44

Now it's accessible to an entire research community. Who can do this secondary use of data, real world approach to? What new questions might I be able to ask if I had hundreds of tracings from hundreds of patients about how CGMs can be useful?

David Klonoff: 19:00

One. Where can people find out more information about the Diabetes Research Hub?

Juan Espinoza: 19:06

Yeah, so our website is up and running. It's with the Diabetes Technology Web, so folks can go to drh for Diabetes Research Hub, drhdiabetestechnologyorg, and there they can learn about how to both view some of the data that we're already hosting, as well as how to submit their own data and how to contact us to get a copy of the local software that they can use on their own data. Thank, you.

David Klonoff: 19:33

Well, this is a really good project and I know that there's more to it than you said in this time frame, and we plan to interview you again later to find out some of the additional capabilities of the Diabetes Research Hub. So for now, on behalf of David and myself, I would like to thank both of you, shahid Shah and Juan Espinosa, for speaking with us about the Diabetes Research Hub. So until our next podcast. I'm going to say goodbye and this podcast is available on Spotify, the Apple Store and the Diabetes Technology Society website. So until our next podcast, thank you and goodbye. Thank you very much. Thank you for having us.

Shahid Shah: 20:20

Thanks.