On this episode of The Huddle, Miguel Johns, Co-Founder & CEO of mmnt and creator of Milton, discusses how the use of artificial intelligence (AI) can benefit health care professionals across the diabetes space and how this technology is growing and changing. Topics include addressing some of the current opportunities and challenges associated with AI in health care, how AI can provide improved clinical decision support, and where this technology may be headed in the future. Explore danatech here: Diabetes technology for healthcare professionals | Danatech (adces.org) For more information on Miguel or mmnt, visit Milton | Powered by MMNT (getmilton.com)
On this episode of The Huddle, Miguel Johns, Co-Founder & CEO of mmnt and creator of Milton, discusses how the use of artificial intelligence (AI) can benefit health care professionals across the diabetes space and how this technology is growing and changing. Topics include addressing some of the current opportunities and challenges associated with AI in health care, how AI can provide improved clinical decision support, and where this technology may be headed in the future.
Explore danatech here: Diabetes technology for healthcare professionals | Danatech (adces.org)
For more information on Miguel or mmnt, visit Milton | Powered by MMNT (getmilton.com)
Dana Moreau
Hello and welcome to ADCES's podcast, “The Huddle: Conversations with the diabetes care team”. In each episode, we speak with guests from across the diabetes care space to bring you perspectives, issues and updates that elevate your role and inform your practice. I'm Dana Moreau, Director of Danatech at the Association of Diabetes Care & Education Specialists. Danatech is an online platform created to address the rapidly evolving diabetes technology landscape. It supports the learning and assessment needs of health care professionals with up-to-date product information, device training, professional education, and more. Visit us at danatech.org.
Now, before we begin, I want to point out that the information in this podcast is for informational purposes only and may not be appropriate or applicable for your individual circumstances. This podcast does not provide medical or professional advice and is not a substitute for consultation with a health care professional. Please consult your health care professional for any medical questions.
Okay, so in today's podcast, we delve into the intersection of AI and diabetes provider enablement, exploring how artificial intelligence is revolutionizing health care delivery in diabetes care management. Our expert guest today is Miguel Johns, co-founder and CEO of mmnt and creator of Milton. Miguel will discuss the role of AI in addressing challenges and opportunities and share some real-world examples of how health care professionals are using it to impact positive patient outcomes. Miguel, welcome to The Huddle.
Miguel Johns
Thanks for having me.
Dana
Glad you're here. Okay, so before we get into today's discussion, Miguel, can you tell our listeners a little bit more about you, your work at mmnt, and really how you got into this type of business?
Miguel
Yeah, so I've always had a passion for diabetes. My dad has type 2 diabetes. My grandmother on my mom's side has type 2. My grandmother on my dad's side passed away from Type 2 complications. So, it's always been something that's been close to my heart. I knew I wanted to spend my life's work working in the space and helping people with diabetes. I started building diabetes technologies around 10 years ago. I got interested in AI around 2016 when IBM Watson first hit the scene, there were a lot of limitations with IBM Watson that have been solved today, which gets me even more excited.
But through that entire process, we've continued to learn, continued to experiment, and now we are in the right place at the right time for everything that's happening. And so, I feel like this is a great opportunity for health care to truly be transformed. A lot of the things that we've been dreaming of are now coming to the surface thanks to these technologies. And just a quick note about mmnt, we provide creative strategy, content, and AI technology to diabetes enterprise companies.
Dana
Terrific. And clearly, and you've kind of alluded to this in your opening, there's a lot of ways that we can go with AI because, I mean, its applications are so broad. So, this first question will be a little bit broad. What really is the role of AI in diabetes provider enablement right now? And how exactly are you seeing it transforming health care delivery in this field?
Miguel
Yeah, so I think where we can focus if we're going to talk about AI and provider enablement is something that was going on before chat GPT took over the world. And that's clinical decision support. Clinical decision support has been around for 10 plus years. It was a lot harder to develop these types of technologies and build AI before chat GPT, before open AI. But the way people used to do that, entrepreneurs used to accomplish that by creating these vast complex decision-making trees.
And these would turn into technologies that would be facing the provider and helping them make the right decision to aid the patient in their care based on a whole lot of data. So, one of the specific use cases is in remote physiological monitoring. If a patient is out in the world tracking their health, glucose levels, meals, medications, things like that, that turns into a whole bunch of data that the provider has to review to make an accurate next step decision for that patient. And even before chat GPT and OpenAI, there was technologies that could help understand that information, help the provider understand it better, and even suggest next steps to that provider so that they could better care for that patient. Now, AI today, with OpenAI really taking over and leading the way, has made that process a thousand times easier. Instead of developing these complex decision-making trees and needing a sophisticated research team and a room and lab full of engineers, you can use natural language. You can type into a website and the platforms underneath all of this are doing the heavy lifting. So that's what gets me most excited is because that use case has already been proven. It's already been around. AI has already been making an impact there. It's now just much easier to create and iterate and develop. So I think we'll see a lot more of that here as we are today and in the near future.
Dana
Thinking about that specifically today, how are health care providers starting to use it and what is it doing or how is it helping so far? What have you seen?
Miguel
The biggest challenge for providers when it comes to patients with diabetes is the follow-through on behavior. And AI is already transforming that piece of the puzzle. So, as I mentioned, to start the clinical decision support, that's one piece of the puzzle. Once the data is collected, AI can understand that data at a much deeper level. And where AI just blows the doors wide open is, now every single human being can have a personalized engagement plan for them specifically based on their life, their health history, their cultural backgrounds, their language, their preferences, their hobbies, all of that is going to come into play and it's going to train these AI models to be extremely personalized, which will lead to better results. So some of the early signs of that that are happening already we're seeing.
Major glucose device companies like Dexcom and Roche and others apply these technologies within their CGM processes to understand what's happening in an individual's trends, day-to-day life, and use AI to make small, smart nudges that can influence that patient to take a better next step, whether it's go on a five-minute walk, whether it's swap out one recipe for another, whether it's take some time for stress management or create a better sleep routine, that is already happening. There was a big announcement just this month actually from Roche talking about their new AI enabled CGM device that will be hitting the market. And so I see AI really making an impact on that end of the spectrum and then bringing it back to that clinical decision support technology so that both ends, both the provider and the patient have a better experience. Which is gonna lead to better patient outcomes for the patient in the end.
Dana
And are you seeing already better patient outcomes from some of the AI applications that you work with or have seen elsewhere?
Miguel
Yeah.
Dana
What specifically might you be seeing?
Miguel
Of course, Selfless Plug, Milton AI, our technology that we've created and now customized for diabetes enterprise companies. We work on the patient engagement side using text messages. We chose text messages because our goal is to increase access. We've all lived through this age and wave of diabetes apps. Most of them do not get used and there's lots of reasons for that, but there's a lot of friction when it comes to those applications. That's why we went the path we went. But to give another organization a shout out and not just put all the spotlight on us, there's a technology created by DioSys, which actually created their clinical decision support AI before OpenAI and Chat GPT were a popular thing. They did it the old way with vast complex behavioral decision-making trees. And DioSisters worked with amazing companies like Cardinal Health, and their technology is amazing at understanding a patient's health history, understanding their condition, how long they've had it, their comorbidities, what medications they're on, and then looking through a range of hundreds of different diabetes therapies and providing them alternatives to the traditional therapies that most everybody is prescribed. So the problem they solve is that a provider is only trained on two or three therapies that they go to to prescribe to a patient, even though there's 100 plus. Isis technology is able to parse through all of that information and make clinical decision support recommendations to that provider and make them aware of alternative therapies that are better for the patient and can save the patient a lot of money. So that technology has been around. They've had over 100 different customers and partners. They produced amazing outcomes. We've been working closely with them over the last about year, and I'm really excited for what they have coming next.
Dana
Now, terrific. Again, obviously there's a lot going on. And it occurred to me as you were talking, and I think we've talked about this before, that I mean, AI is a tool, not necessarily a replacement. What really is the role of the health care provider in working with these tools and working with people with diabetes? What does that trio look like, I guess?
Miguel
Yeah, you're exactly right. Everybody, when AI first hit the scene across industries, the first question is, is AI going to take my job? I try to make it very clear to health care providers that AI alone is not going to take your job, but health care providers leveraging AI will greatly outperform the ones who do not. Now, one of the areas where we see these tools being leveraged in the best possible way is just in a greater understanding of what's happening in the patient's life. So unfortunately, in most situations when a provider and patient are able to get together, there's so much context and information and data that's just left out because of many reasons, but a big one being time. The provider can't spend all day every week, every month with the patient, which is where the patient is living their life. They're living 99% of their life outside of that interaction with the provider. And so AI's potential, starting out the low-hanging fruit, is to be able to gather much more of that context, a lot more of that data and those insights, and make it extremely simple and consumable for that provider. That goes back to that clinical decision support technology that I started the conversation around.
because AI is going to bring all this data, but all this data is useless if it's just thrown at the provider who's already too busy, already overwhelmed, already stressed out. It's got to be simple. It's got to be the most important things at the top of the list. And that's where I think this is going to make the biggest impact. Now, what was really interesting to me, I was recently writing an article for the diabetes practice group for the Academy of Nutrition, and there's a lot of research going on around how AI can dive into the data within electronic health records. And the idea being, hey, if we can leverage AI within these health records, we can identify different social determinants of health and provide better care to these patients. Well, the barrier currently is that these electronic health records are mostly powered by the clinical notes, which are very sparce. There's gaps, there's missing context, they're written quickly, they're only about the most important thing in regards to the medications or the treatment plan.
And so there's so much information missing from these clinical notes. So the result of those research studies was that we need more information within the electronic health record for this to really be effective. Now, take it to the other side. The number one use case we're seeing across health care in general is clinical documentation. So there's a startup called Deep Scribe that's becoming really powerful. Of course, we're competing with electronic health record companies like Epic and others who are also taking this action step and rapidly rolling out clinical documentation tools because one, it's the biggest pain in the butt for health care providers. They don't want to sit there and document every single day and chart every single day. But two, this is going to bring so much more context into the picture. And then these technologies that are in the electronic health records and going through this data, they're going to become far more effective because they've just got so much more information.
So whether you're in health care or in any industry, the name of the game is data in, results out. And these AI technologies are useless unless they're getting valuable data coming into them. And I'm seeing that use case play out when it comes to clinical documentation, which is gonna empower these providers and these health care teams to deliver better care to the patient.
Dana
Exactly, I'm noticing a primary marketing message being that a lot of these new tools are HIPAA compliant, which was always one of the primary concerns. When we're thinking about AI also, in what ways re you seeing AI enhance the efficiency or effectiveness of the diabetes diagnosis and or monitoring compared to some of the more traditional methods?
Miguel
So monitoring is is right in our wheelhouse again because of Milton AI. That is an area where it's real time. It's in the moment for the patient and their life and helping them track the important information like meals, like glucose levels, stress factors, obstacles, interests. There's so much context to an individual's life that plays a factor in their health. And I think AI is the tool, and we haven't had a tool beforehand that could accomplish something like this. Now, dietitians and CDCES's have been doing this already for years. They've been engaging with patients in between visits. Some of the best programs that have been the most effective have used text messages to collect insights and patients would send in pictures of their meals and the dietitian would review that. But again, there's a limitation in the amount of health care providers in comparison to the rapidly growing number of people with diabetes. And so AI has the ability and is doing so today.
We know through firsthand experience to engage that patient, just like a dietitian or CDCES would collect that data and make it consumable for that health care professional. That's just one area. I think when it comes to diagnosis, one of the biggest issues in the health care space today is there's a whole lot of people with undiagnosed diabetes. And AI will make it possible to understand based on what somebody's health history is, based on prompting them with specific questions at the right moment to be able to diagnose people even outside of the clinic, even without necessarily an A1C test to start with surveys and with questions.
Dana
You said something earlier about the data and obviously the recommendations that you'll get out of these tools are only going to be as good as the data that is fed into the tool. So in the past, not just with AI tools, but with other tools, it relied a lot on self-reporting, which people may start out strong, but sometimes that tapers off. I'm wondering, like right now with tools that are existing, maybe like Milton and others, are they grabbing data from everywhere? Are they taking the AID data? Are they taking the CGM data? Are these integrated now with all the tools that are existing or emerging? I guess, where is that today and where is it going?
Miguel
Yeah, so we're already seeing the beginning of that happening, whether it's the major CGM companies applying AI and their technologies directly. I know we're working with a couple of companies who do a very good job at integrating device data and we're partnering with them to bring that into Milton. Now what we don't want to do is overstep our boundaries because that's going to be a quick turnoff for the end user. We don't want to just scan their phone and their messages and Milton actually can't do any of that because it lives within the text message world. And that's been a problem in the past is privacy concerns.
Now, where I think the opportunity is to overcome that obstacle is in the extreme personalization. AI will learn about the individual. It'll take input from the individual based on their preferences, their hobbies, what they do on a daily basis. And with that information, the messages to re-engage the patient will become far more personalized. The issue in the old world is, I might be able to create a re-engagement campaign that's going to go out via notification to all of my app users, but there's only so many variations I can create and they're still going to be very generic. They're not going to be based on that person's specific needs. And that's where AI is an amazing tool. It can solve that problem. It can understand the person better. And so when that person begins to disengage or fall off, it can hit them with the right message that re-sparks that motivation and that inspiration really drives them towards the reason they started in the first place and gets them back going. Now, the jury is still out on that. There's lots of research still to do that these tools are still very new. But that is where I'm very passionate of the possibilities is that that personalization to re-engage, re-inspire, re-motivate people when they do inevitably start to fall off.
Dana
What advice would you give to health care professionals looking to leverage AI for diabetes enablement? If there was one thing, one piece of advice, what would you give them?
Miguel
Yeah, you need to find an expert. So the largest organizations are now hiring prompt engineers. They're hiring internal AI teams. They're taking the action steps to have experts shoulder to shoulder with the clinical team, with the operations team, with the marketing teams. That's absolutely key. Play with the tools, of course. Hop into chat GPT, hop into Dolly, make images, make patient educational materials, get started, find people online that you can follow and learn from, watch YouTube videos if that's your means of communication. But you need an expert if you're gonna really roll these out in your organization internally or work with a company like us at mmnt. That is step number one because you'll quickly learn that there's a lot of unknowns when you first dip your toe in and you'll end up wasting a whole lot of money and wasting a whole lot of time and might not get anywhere really fast without that expertise. So start there, teach yourself, play, play with these tools. But if you're going to really take action and apply these into your business, find an expert.
Dana
Good advice. Is there anything else that we haven't covered that you need our listeners to know?
Miguel
While all this information is really exciting, AI is definitely the topic of discussion right now. And there's so many possibilities. There are also challenges. We need to be thinking and considering the regulation implications, the reimbursement implications, data security, HIPAA compliance is absolutely necessary with these tools. Let's keep that top of mind. Let's have clinicians involved in that discussion, not only technologists and policymakers, but clinicians as well. That's something that we all have to figure out together and it's got to take time and there's got to be hiccups for sure. But let's remember that as we build on the excitement of AI.
Dana
That's terrific. Miguel, thank you so much for joining us today.
Miguel
Thank you for having me.
Dana
Thank you for listening to this episode of The Huddle. Make sure to download the resources discussed on today's episode. You can find them linked in the show notes. And remember, being an ADCES member gets you access to many resources, education, and networking opportunities. Learn about the many benefits of ADCES membership adces.org/join. Finally, for more information on Miguel or mmnt or Milton, visit getmilton.com. Thanks so much for joining us.