We're going to go into our panel discussion now. I'd just like to introduce our other panelists, and we have a few questions that have already come in. We have Shuk Chenderu, who serves as Chief of Data and Analytics Officer for Anthem, leading enterprise data management, data quality, data governance, and enterprise analytics and applications. We have Dr. Cadman Leggett, is a clinical gastroenterologist at Mayo Clinic in Rochester. He's an active member of the AI and GI leadership group at Mayo Clinic, where he leads initiatives focusing on designs and implementation of AI solutions for endoscopy. And lastly, we have Dr. Helmut Messman, who's the head of the Department of Gastroenterology and professor at University of Augsburg. He's the current president of the ESGE. So welcome, everybody. And let's start off with some questions. So Srivanthi, I really enjoyed your talk. There was a question that came in that raises the question of, how do we inform patients that we're going to use their information for AI research and data mining? Should we start putting this in the standard consent, or should we do a separate consent form? Well, that's an interesting question. So I think it all depends on, one, your institution and the regulation they have. So when you are enrolling patients for research and using their data for research or generating a commercial product, you need to specify the reason why you're collecting that information, because they can be different pathways. So definitely talk to your institutional review board about that consent. From our institution standpoint, we do have separate consents if we are going to use information for a research standpoint as opposed to routine quality of care and approved measures. Thank you very much. Ashok, a question for you. You've seen some amazing technology today of where AI is headed in the field of endoscopy. How do you envision, from a payer perspective, physicians and hospitals getting reimbursed for incorporating this type of technology? Yeah. Thank you, Seth. I'm really glad to be here. As a payer, we collect an enormous amount of data, the data we typically get from claims. We are also collecting medical records, data including pathology, histology, data, all the images, and being able to mine the images. And one of the key focus as a result of collecting this data is, how do we simplify the prior authorization process, or how do we come up with innovative payment models that incentivizes the provider to focus on the quality of care versus the volume? So in a nutshell, our focus is simplifying the admin burden, being able to leverage this data, and combining it with some of the device data, with genomic data, and also social drivers. Like food transportation, and housing, behavioral health, psychographic data, and being able to generate the right insights, both from structured and unstructured. And the reason why, because of the high compute power, graphical processing units, we're able to crunch all this data and generate these meaningful insights towards improving the quality, but also simplifying the admin burden for the providers. Thank you. And speaking of that, Kadmin, you heard David's talk, really reviewing for us all the quality metrics and things that we have to have just to do a basic, say, colonoscopy. What are some of the things that your team is working on in this area to improve the overall endoscopy physician? Of course, the patient's always in mind, but what about for the physician and nursing team, improving their experience? Yeah, fantastic question. Thank you. And thank you to the course directors for the opportunity to be part of this conversation. If I may, I just wanted to give a couple of shout outs to two members of the audience. Jock Bergman is joining us from Amsterdam. He's a leader in the field in terms of Barrett Esophagus, AI development for CADD, CADDX algorithms. And Dr. Nina Coelho, who is from my institution back in Mayo Clinic in Rochester, one of the rising stars in the application of IBD and AI. To answer your question, I was fascinated by your talk, Dr. Armstrong, regarding the smart endoscopy suite. Yes, there are multiple moving pieces as AI becomes more and more common in our practice. We think about AI as something that we can just plug and play, that it will incorporate seamlessly into our current endoscopy suites, but there's more to that. As the field evolves, we are moving very quickly to video data acquisition. Right now, most endoscopy suites are set up for video acquisition, but that's performed still in a manual format. I think automated video acquisition is going to be a key part of data collection, how AI will be applied not only in real time, but also asynchronously to our endoscopy suites. With regards to the application of AI in real time, so that's most of the algorithms that have been discussed in the conference, CADI, CADX algorithms, these require IoT edge devices to be installed in the endoscopy suite. And every endoscopy suite is a little bit different, but things that we are currently looking at and working on is, how are these devices going to be incorporated into our daily practice? Are we going to use a second monitor to display the AI algorithm, things like that, so that they don't interfere with your regular workflow? We find that some physicians are embracing the field of AI and what it brings, and other physicians may be a little bit more reluctant. And I think that the less these algorithms interfere with the current workflow, the better. Great, thank you. David, next question is for you. Sort of what was mentioned from Shivanthi, one of the folks that joined us today is raising the question of, do you think that reporting bowel prep quality and percent of mucosa examined will become a standard feature of artificial intelligence during endoscopy and colonoscopy? And will that move away some of our other quality metrics that we're currently using, like withdrawal time? Thank you. I would agree with that, and I think one of the challenges that we've had over the years has been the sort of assumption or hope that there is going to be a single metric that will improve our sort of performance as endoscopists. And actually, when any complex sort of visual motor task that we do has a number of parameters that we monitor as we go along. So I don't think there's going to be a sort of holy grail or single sort of metric that will deal with this, but they are clearly important. And so things like bowel prep and percent of mucosa visualized are sort of intuitively really important parts of what we do. We need ways of being able to assess that and not necessarily to sort of document whether or not we're good or bad endoscopist. I think part of this is giving people feedback so that they can improve their practice and they got a way of incorporating feedback into their practice. So I would envisage, if you like, in response to the question, having a suite of performance indicators that provide feedback in a manner that allows endoscopists to improve that performance to monitor what they're doing and then to know where the gaps are that they can address. And Helmut, question for you. President of the ESGE, physicians really rely on societies for guidance in terms of changing practice to move towards best practice and also to implement technology. And so right now, artificial intelligence, what we're seeing that's commercially available is for polyp detection. What do you think from a society level we will need for the society to recommend physicians should really adapt this? Because when you bring the notion of we're going to improve your polyp detection ADR, some physicians say, well, I have a pretty high ADR. How much higher am I going to get? And what's the value add? So thank you for this difficult question. I mean, ESGE tried to give some recommendation on quality parameters, which are definitely important for our daily practice. And we saw two fantastic lectures today showing that AI can improve our quality. At the moment, I think we get some experience showing that AI works in clinical practice. We have first experience showing the detection rate for small polyps mainly increases. But how will it influence in the near future our daily work? How will it influence, for example, our fellows? How will it influence safety and other important things? And we did a first recommendation in our guidance that, well, AI can be used. But it's up to now. I think we need data. We need data. We need randomized trials to be sure that our recommendation as a society, and the same as for ESGE, if we say, well, for daily practice, AI colonoscopy is recommended. So this is an issue that it can be a legal point. Now, if you don't have AI, for example, and you will miss a polyp, that can be for sure a problem in the future. Right now, we don't have this recommendation. I would say work in progress. But in a couple of years, I think many societies, the German one, even the European, and probably I don't know what the American society will recommend. But I think AI will definitely influence our daily practice and even those legal aspects as well. All right. Thank you. One last question to Ashok. How do you, as an insurance provider, use AI for quality assurance for your members? What are the things that you're looking at? And what do we need to do for AI to be fully integrated on a daily level for clinical practice? Yeah, absolutely. Great question, Seth. So I would say a couple of things. One is, and I think a lot of the panel members and even in the talks earlier, it's all about the data. And there is a lot of data. How do we discern the right data and remove the noise? So when we get the medical records, or the image analysis, or the pathology histology data, and combine all of that, and then use that to influence whether it is the clinical guidelines, or whether it's being able to calculate the quality measures on a timely basis, and even share that back into the providers and integrate it into their workflow. We have partnerships with a lot of large providers and physician practices, because we can generate a lot of AI, but the key is making sure that information gets integrated into the workflow, and also being able to share with the members as part of being able to notify them, or if they're leveraging digital tools, whether it's MyChart, or we built our own digital tool as well, so that way they know what action to take. So in a nutshell, it's about the data. It's about discerning that data to generate meaningful and timely quality insights, and then taking those and sharing it with the providers and with the members in their workflow. Great. Thank you. I want to thank Philip, my co-moderator, our speakers for this section, and our panelists. I'm going to turn it back over to Dr. Wallace. Thanks, everybody.

AI in healthcare

improving endoscopy procedures

data use

clinical practice

quality of care