Advertisement
Nutrition411: The Podcast Ep. 35

The Use of Artificial Intelligence to Improve Diabetes Care, Education

Lisa Jones, MA, RDN, LDN, FAND

In this podcast episode, Lisa Jones, MA, RDN, LDN, FAND, interviews Rachel Stahl Salzman, MS, RDN, CDN, CDCES, and Livleen Gill, MBA, RDN, LDN, FAND, on ways artificial intelligence can be used to improve diabetes care and education, including challenges in the use of AI and resources to stay up to date on new and up-and-coming diabetes technology.

Additional Resources:

 

TRANSCRIPTION:

Host: Hello and welcome to Nutrition411: The Podcast, a special podcast series led by registered dietitian and nutritionist, Lisa Jones. The views of the speakers are their own and do not reflect the views of their respective institutions or Consultant360.

Lisa Jones: Hello and welcome to Nutrition411: The Podcast where we communicate the information that you need to know now about the science, psychology, and strategies behind the practice of dietetics. Today's podcast is part of a series of episodes on diabetes technology featuring Q&A with Livleen Gill and Rachel Stahl Salzman. Welcome Livleen and Rachel, thanks so much for having us. I'm excited about this episode. I want to take a moment to introduce each of you first. So, I will start with Livleen Gill. Livleen is president and CEO of Apostle Group, LLC, a consulting company that provides innovative solutions to clients in healthcare, food, and nutrition. She's also the CEO of Wholesome Village Inc. in Rockville, Maryland. Previously, she was a private practice nutrition consultant for more than 20 years, and the Food and Nutrition Services Director and outpatient dietitian at healthcare centers in Maryland. She will serve as the Academy of Nutrition and Dietetics President in 2024-2025 year. Next, I want to introduce Rachel. Rachel is a registered dietitian and diabetes care and education specialist in the Division of Endocrinology, Diabetes, and Metabolism. At Weill Cornell Medicine in New York City. Rachel is passionate about empowering individuals to make sustainable lifestyle changes and leverage diabetes technology and digital health to improve their health and quality of life.

Alright, well thank you both Rachel and Livlean for joining me for our discussion today on artificial intelligence. And it seems like this is a topic that everyone is discussing. You can't go a day without somebody talking about how they were on Chat GPT or so. There are so many things coming out. It's hard to keep up with everything. So my question is, how can AI be used to improve diabetes care and education?

Rachel Stahl Salzman: This is absolutely such an exciting field that we're continuing to see evolve and evolve, and so I think it has many different use cases that can be applied. And keep in mind that AI has been used and is still being used in a lot of diabetes technology already. So, I think some areas, and Livleen could add here, where it can be used are certainly in the nutrition realm. We could use it for meal planning support to help inform the people we care for, how to eat healthfully, or how to understand their meal plan if they have certain allergies or or dietary restrictions. And we could use these large language models like Chat GBT to help us design it. And I'm sure it's only a matter of time, if they might already exist, like specific nutrition-related ones that are able to pull the appropriate data sets to make it as accurate as possible.

And I'm sure we'll talk about some of those concerns there. So I think the nutrition side can be really helpful and something that I use very often in helping with that. Another area that I've been following is AI in screening and diagnostics and diabetes, and specifically in the realm of AI being used to help detect eye issues. So there's been a lot of published research on AI and retinopathy in an AI system, being able to identify a form of diabetic retinopathy in a person. And this has so many amazing implications because there's a shortage of eye doctors to review it, and this is helping, in so many ways, improve efficiencies. So, those would be just two short examples, and we'll see where Livleen goes from here.

Livleen Gill: I think Rachel touched upon a whole lot. So, from the nutrition side, absolutely the diet monitoring, the education piece of it, and how to use that, how do we predict and personalize that for the individual? I think that would be where AI would be a big help. But I think the machine learning part of AI, which is now what AI is, I think is the piece that we'll know more about as it becomes more accurate to predict and help with personalization of the services. The other thing is what Rachel alluded to with retinopathy, and I'll say it's the population kind of screening and being able to predict who is going to develop certain conditions or diseases. I think with AI's predictor model that is where it starts to help clinically, I think it would really do wonders for having a shortage of providers. I think AI helping that would be very crucial. So, those are my data points.

Lisa Jones: And they're excellent. There are just so many advances. I know some companies are doing this already, like in the pharmaceutical industry, their information is proprietary. So, they've created their own gen AI that they're using, basically, you can upload, whether it's the PI for the drug or whatever they're working on, their employees have to sign off on a policy that they won't put the information someplace else, like in the regular Chat GPT. Are you aware if that is coming to a hospital setting?

Rachel Stahl Salzman: That's interesting. Yeah, I think we need to be careful with data privacy, so I'm not surprised. I think that is definitely an area that we need to be careful with, for sure.

Lisa Jones: Yeah, I'm just curious. Like I'm thinking if the pharma industry has started it, the hospitals are probably working on it. I was just curious if anybody's heard anything.

Rachel Stahl Salzman: Another benefit that I think could be really helpful and that we are starting to see in help with diabetes care and overall registered dietitian efficiencies, it is my hope that AI is giving us the gift of time, right? I think that we're going to see AI helping with a lot of the administrative work. Maybe it could analyze a patient's intake form, providing us at the time of the visit with some important key points to consider in our discussion. I think when I look at a patient's food log on one app and their CGM data on another and maybe their pump data, how could AI kind of put all those pieces together from all those different software and platforms into a way that can help summarize and guide me in helping support them, like Livleen said, ultimately provide more personalized recommendations to help in their treatment plan.

Lisa Jones: Yes, and I love what you said, "the gift of time," and I do want to add to it that the gift of time, AI does save time. Let's talk a little bit about the challenges. So, there are some challenges that need to be addressed before artificial intelligence can be fully integrated into diabetes care and education. Do you want to talk a little bit about what some of those challenges are? Let's start with Livleen for this one.

Livleen Gill: Rachel alluded to one of the biggest ones–it's regarding data privacy–that the data that is collected...what are the things that we put in place to make sure that that data for those individuals is private and HIPAA compliant. Then the second thing is the integration of AI into the clinical and digital workflows. Again, Rachel alluded to that. How does it take all of this and in different disparate forms that we are getting, how does it come together for us to be able to make sense and provide the care that we need to for the patients? And the third is legal challenges–medical malpractice–how are the legal challenges, how is that going to get incorporated into this and what are the implications of that? So, those are the three challenges that I see in incorporating AI into practice.

Lisa Jones: I definitely agree with all three of them. But just as we've seen over time, since AI rolled out, the advances in it, like that example I was using within the pharma industry, I think that's how they are overcoming some of these challenges. I think we're going to get there. It's just going to take time. So really great points. Thank you, Livleen. How about you, Rachel? What are your thoughts on some of these challenges?

Rachel Stahl Salzman: I love those three, Livleen. It's really important for dietitians who are listening to understand those different challenges. And I have three more, so we've got six total to add here. I think about, in the ways that I've practiced using some of these large language models, this idea of hallucinations. The idea that these large language models are producing output that might not exist. It could be incorrect or totally misleading. So, therefore the human eye and really understanding and verifying the information that's coming from the outputs of these systems really needs to be looked at. For example, I was practicing using it to help with a literature review for a research study. And I noticed that for some of the information it provided, it cited a research study that didn't exist. Or at least, in all my searching, and I feel like I have pretty good search skills, I could not find the article that it cited some information from. So, we definitely need to be careful of that. The second one is around data quality. So, I think I mentioned earlier, that these AI systems are built on specific data sets and so are these comprehensive? Are they diverse? We don't fully know and therefore it could lead to incomplete or inaccurate data, which could then lead to a person maybe taking information that's flawed from the system. And then the third one is around the idea of biases that these AI systems could have biases built in and these biases could be from several different sources, one of which could be unconscious. Maybe someone's building their own specific AI system and the founder, based on his or her own experiences or his or her own research, is putting in data to train the system that might not be truly comprehensive. So, that could all lead to inconsistent and potentially biased results.

Lisa Jones: Some of your points prove the need for the actual human RD to review the output of what the prompts are generating. I've seen that before too. I put something in, I've never seen that study, and then I had to go try to find the study. I couldn't find it, and I was like, well that's not even factual.

With all of this technology, how do you stay up to date with everything surrounding diabetes technology? Because it just seems so overwhelming.

Rachel Stahl Salzman: I completely agree that it can be really hard to keep up with all things technology, but the good news is that there are a lot of resources to help. I'll share a couple here. One of which again, we're with a group of registered dietitians, is joining the diabetes DPG. This is one of the largest dietetic practice groups of the Academy of Nutrition and Dietetics. As the chair of the Diabetes Technology Committee, my goal is to help make this information accessible; making this information easy to understand, with the ability to access it in multiple forms. We have a website which, even if you're not a member, you can go to and access our technology website. We also have webinars. We also produce different publications, one of which is coming from our cutting-edge nutrition and diabetes care publication later this year. So there'll be a lot of information on AI.

Additionally, as Livleen might agree, going to conferences could be a great way to stay up to date. I love the ADCES annual conference and they have a dedicated separate conference just on diabetes technology where you can get your hands on some of these devices and certainly participate in so many different sessions available for you on that. And then online websites, like the Panther Program, which is available at pantherprogram.org, has a variety of clinical tools related to diabetes technology. They have a great insulin pump comparison chart, which I love and I share that with patients when they're talking about and interested in different pumps. We share it with other providers and they are constantly being updated. And then also danatech.org. That's through ADCES where you could search different CGMs, insulin pumps, and a variety of other diabetes technology tools.

Lisa Jones: Oh wow. That's very comprehensive, but they sound like excellent resources. And then I'm wondering too, just a quick follow-up question for you, Rachel, do you think stuff is happening where you have to update this every month type thing? Is it happening that quickly? Or how quickly would you say things are coming out that you have to stay on top of?

Rachel Stahl Salzman: It could be monthly, but you never know with the field of diabetes, which certainly keeps it such an exciting field. But again, I keep these websites saved on my browser, so periodically, you can easily check it for the latest updates. These resources are already pulling the most updated information so you don't have to be searching in the blind web for this information.

Lisa Jones: That's really helpful to know, especially for those dietitians who are up and coming trying to get into this field that feel like they have this really big learning curve where it's like, "Oh my goodness, how am I going to keep up with all this stuff?" You're saying if you have the sites bookmarked, it's much easier. So thank you for that. Thank you for sharing all those resources. And then Liveleen, what are your thoughts? How, how do you stay up to date on all these things diabetes technology?

Livleen Gill: Well, for me, number one is hands-on, which Rachel alluded to, going to conferences. So, I like to dedicate my time, because during my work day, I don't get the time to follow up on these things. So, I make sure at least once a year, I attend a live conference. The second one, I would say The Academy of Nutrition and Dietetics, DPG, the diabetes DPG, they do a really good job. And the magazine too. Browse through and get a little update to see what's coming and what's the latest. And the third one where I've done certain certificates is through ADCES, through danatech. That's also a great place to do it. And then, some of the journals that I get and the different reps that come through, you also learn from them what's coming down the pike. Those have been, mostly for me, my way of learning about new technology.

Lisa Jones: Thank you, Livleen. Thank you for sharing that. I think all those examples of how you're staying up to date are great for others to hear too. And what I've learned from both of you and I think my biggest takeaway is you never stop learning and there's always room to improve. And that's what all these resources show us. That's what AI is teaching us. So, thank you for sharing that. And I do want to ask you one final question and that is, what is your biggest takeaway from our AI discussion today?

Livleen Gill: I don't think we should be scared of AI or that it's going to take away our job. I think what Rachel said, it's a gift of time. It's going to improve efficiency, it's going to improve the care we provide, and it will become more personalized care. It is just another way for us to improve the care that we deliver.

Lisa Jones: Yes, I love that. And I also think of it like it's the person sitting next to me, even though it's artificial intelligence, that's helping me. It's like having another dietitian that's helping work alongside you. And what do you think, Rachel?

Rachel Stahl Salzman: Yeah, I completely agree. AI has huge implications across health and medicine and it's really made us start to rethink and reimagine how we practice. I encourage all the listeners to try it out, practice, and read up about it because it's here to stay. And as all of us are sharing here, it's not there to replace us, but it can only be there to augment the care that we provide. So I'm excited to see as it continues to develop and see where it goes to ultimately support the people we care for.

Lisa Jones: Yes, and that's the main goal, right? To support the people that we care for. So thank you. Thank you both. Thank you, Livleen and Rachel for sharing your wisdom with us today on the subject of artificial intelligence and diabetes. Thank you.

Host: For more nutrition content, visit consultant360.com.

© 2023 HMP Global. All Rights Reserved.
Any views and opinions expressed are those of the author(s) and/or participants and do not necessarily reflect the views, policy, or position of Consultant360 or HMP Global, their employees, and affiliates.