Every minute, our dedicated AASLD community is leading the way for the prevention, treatment, and cure of liver diseases. From the beginning, AASLD has been at the forefront of advancing the field of hepatology and best patient outcomes through top-tier educational programs, global collaborations, and supporting critical research development. AASLD is at the heart of an ever-growing web of connections that bring together cutting-edge science, groundbreaking technology, life-saving innovation, and the community of incredible minds that make it possible. We foster groundbreaking research and collaborations that lead to improved treatment options for millions suffering with liver disease worldwide. We are a vibrant alliance of mentors who support and encourage one another. Our day-to-day victories pave the way for a healthier world. We are clinicians, educators, researchers, and allied healthcare professionals on a shared mission driven by science, connected by community. We are AASLD, moving the mission forward every single minute. Welcome to the Liver Meeting Digital Experience 2021. And now, AASLD President, Raymond T. Chung. Hello everyone. I'm thrilled to welcome you to this year's Liver Meeting Digital Experience. When the year began, we had harbored such great hope of reuniting in California to celebrate the return of TLM. Indeed, by spring there had been promising signs that COVID-19 vaccines could effectively put the pandemic behind us. Unfortunately, the recent surge of the Delta variant forced a change of course, since we could not conscionably risk hosting an in-person event. So I welcome you back, virtually, for a second year. While we are disappointed that we will not see each other in the flesh, I want to emphasize that TLM DX21 will be far from just another digital meeting. We know that there is so much more to the Liver Meeting than just the science, and our science this year is superb. So we challenged ourselves to deliver a maximally interactive experience. This year, our theme is Driven by Science and Connected by Community. To this end, you'll find new ways to explore and connect with your vibrant community within this year's platform. You will discover new avenues for exploring posters and engaging with poster presenters. We have incorporated new technologies to make spontaneous connections with your peers easier and more entertaining. Throughout the meeting, there will be a variety of special interest group networking sessions. You'll also find networking opportunities for early career members and trainees, international members, patients, and exchanges with our friends at NIH and other federal agencies. These networking events will only be happening live, so please be sure to check your schedule and block off your calendar. Also each night, please join us for social events to cap a day of learning with fun, entertainment, and even more networking. We have exciting activities on tap for Friday, Saturday, and Sunday evenings. The pandemic has taught us a lot about our community and what we want to achieve as a society. It has accelerated our move to become a year-round beehive of programming and interactivity rather than a monolithic, single-meeting-centric society. As disappointed as we were to lose the chance to come back together in person, we are excited to announce the return of a series of smaller in-person meetings in 2022, including several special interest group-driven Emerging Topic Conferences and a NAFLD nomenclature Consensus Conference, on top of many more virtual Emerging Topic Conference programs as well. We look forward to hosting these smaller meetings even as we prepare to gather again at the 2022 Liver Meeting in Washington, D.C. Mentoring across the career lifecycle is a critical priority for AASLD. As our world has become increasingly digital, we've expanded the reach of our mentoring programs by bringing several of our committees together in a joint initiative to assure that our members can continually access AASLD mentors across a virtual network. In fact, special mentoring sessions have been planned during this year's meeting. Please keep an eye out for them. We've also kicked off an AASLD Thought Leader Series to have many of our experts share their stories on how they built their clinical and research centers of excellence. These member-driven initiatives are not only welcomed by AASLD, but hugely encouraged. Your members are our greatest strength, and we want to harness your creativity, insights, and initiative to bring to our community the tools, resources, and programs that you want and you need. Another major area of focus and pride for this society have been our diversity, equity, and inclusion initiatives. We remain as committed as ever to ensuring we do our part to tackle healthcare disparities and inequity in our field. Under the guidance of our Inclusion and Diversity Committee, we are recognizing meritorious healthcare disparities research abstracts and assuring balance of underrepresented minorities and women in our leadership and our committees. We are also planning initiatives to enhance the pipeline of underrepresented minorities to hepatology and to mentor those who are already in our field. Our connections with our global partners have only been strengthened during the pandemic. With our friends at Global Liver Societies, we have not only continued our highly successful joint symposia, research workshops, and Connect educational programs, but we have also launched hepatitis elimination and liver cancer programming with our partners and forged brand new affiliations with our friends in Africa. Our partnership with patients has grown dramatically, and we have doubled our programming for patients at TLMDx and beyond. We're also proud to have hosted a congressional briefing on NAFLD and continue to lobby for passage of the NASH CARE Act with our patient partners. With the help of our members and committees, we are demonstrating leadership in other domains as well. For instance, we are preparing to launch a universal application and set community standards for transplant hepatology fellowship programs. We have also endorsed and are supporting the Secure Liver Registry, a highly valuable database of COVID-19 cases and vaccine breakthroughs in liver disease patients and transplant recipients. We do all of this while maintaining a laser-like focus on that which drives us, finding cures for liver disease. Even during a challenging time, your continued commitment to high-quality science has been astounding and is reflected by the over 2,000 abstracts being presented at this meeting. And despite restrictions and travel, our community has continued to deliver high-quality liver care to hard-hit and underserved communities. You have remained committed and taken the time to join us to keep learning and growing. So this year, we salute and thank you for your dedication to making our community incredibly strong. Putting together TLMDx21 and our programming throughout the year has been made possible only through the tireless commitment of so many in the Association. A special thank you to the co-chairs of the Scientific Program Committee, Drs. Lori DeLev and Meena Bansal, and to each of our Scientific Program Committee members. And of course, a giant thanks to our ASLD staff, who have spent countless hours bringing this wonderful meeting together. We hope you have a great TLMDx, and we look forward to engaging with you. With the explosion of digital data in medicine, the power and potential of artificial intelligence approaches to optimize health care outcomes is immense, and there could be no better person to speak to these possibilities than this year's President's Choice Lecturer, Dr. John Halamka of the Mayo Clinic. Dr. Halamka is Professor and President of the Mayo Clinic Platform, an incubator devoted to tech and data innovations in health care. He has helped transform the way we think about the use of machine learning for patient care. So without further ado, Dr. Halamka. Well, it's a pleasure to speak with you today on the future of AI in medicine and hepatology. My name is John Halamka. I'm President of Mayo Clinic Platform. I will assert that over the next six quarters, and note I didn't say five years because that's too long, we're going to see technology advancements, we're going to see policy and regulatory change, and cultural expectations that will ask us to deliver cures in novel settings using novel methods and processes that will require us as providers to rethink the way that health care works in this country and internationally. To start with, let me ground you in the notion of what is a platform, because this is going to frame up the discussion to follow. This may look like a tractor to you. And John Deere, well, hey, as a tractor company, they buy tires, they assemble them, and they sell you a tractor and done, right? Well, actually, five years ago, John Deere became a data company. They've created the world's largest precision agriculture platform. Every one of these tractors is a sensor. It has GPS. We understand what you are harvesting, how you are harvesting, and even the quality of what it is you're harvesting based on the sensors. So with that precision agricultural database, an individual farmer can actually understand how to achieve a good outcome, and a population can understand yields season to season, even commodity prices. Sounds a lot like medicine, doesn't it? And so if we were to become a platform of data transformed into wisdom, what are some of the things that we are going to need to do? I think we would all agree as whether you're a provider or whether you are just a care navigator for a family. I am almost 60, and I navigate the care of those younger and older and my peers. In 2020, 2021, care is often challenging to coordinate. It's not clear where you go next, what disease state you have. Bringing the right patient to the right facility to the right care at the right time at the right cost is guesswork and often requires consultation with friends who may very well have expertise to navigate the system. I think we want something different. I hope by 2030, we want continuous care that's easy to access and navigate based on evidence and make this care equitable and highly available to all. And if we're going to achieve that, there are several prerequisites, and let's go through some of those. The way Mayo Clinic has thought about the work ahead is in four products or four categories. The first, we need to be able to gather data, novel sources of data, not only the standard electronic health records, structured and unstructured data, but telemetry, digital pathology images, and omics. We need to ingest and curate that data. We need to separate signal from noise. We need to unify the data across an entire patient experience and not just a single episode of care. So the first work Mayo has done over these last 18 months is to create a gather function. We've actually spun out a company that is a joint venture with Mayo called Lusum Health that functions around these novel, not standard data sources and bring them into our cloud so they can be used for AI modeling. And having wonderful data, that's really prerequisite, but turning that data into wisdom is hard. What has Mayo done? We moved 60 petabytes, 10 million longitudinal patient lives into Google Cloud in a de-identified container we call data behind glass. The idea is, is that not only is it de-identified, but we control who uses it, how it's used. The data cannot be exfiltrated from Mayo, nor can data be brought in from outside Mayo to potentially link and re-identify patients. But this corpus of data, which is structured unstructured telemetry images and omics is used to create an AI factory and that AI factory turns out algorithms in a variety of specialties, including liver disease. And once we have those algorithms, here's a challenge. How do we ensure they're fit for purpose? How do we ensure that they're equitable and fair? We need a validation function. Taking a million cases from Mayo Clinic in Rochester, Minnesota, and creating an algorithm that is then used in Spain might not work so well. So Mayo has, again, partnered with industry to think of ways to validate these algorithms, to label them, to ensure there's transparency about how they were developed and how they'll be used. And once one has these validated algorithms, you need to deliver them in workflow to providers, to patients, to pharma, to payers, to employers, wherever they're going to enable an action. And so let's go through a few exemplars of the kind of things that you would generate using gather, discover, validate, and deliver. So in the fields of hepatology, I think we all agree. Some of the AI algorithms are going to be based on looking at laboratory results and their evolution over time. Some are going to be based on phenotype and genotype information. Some are going to be based on imaging analysis, and those could be CT or MRI images, or they could be fine microscopy in digital pathology images. And I show you on the slide a whole variety of areas in the field of hepatology based on the literature that have already generated interesting algorithms. And those algorithms at the moment are very much in a research phase, but need to be moved into standard workflow, standard clinical decision support, and be part of our day-to-day practice. I'm going to give you an interesting example, which is not listed on this slide, and it's just so timely, because just moments ago, and I'm being completely honest here, I was just sent this photograph because I am the world's expert on poisonous mushrooms and plants. And just a few hours ago, a 63-year-old woman made a very large meal with this particular mushroom, which some of you may recognize as Amanita phalloides. What is going to happen to this woman typically is that based on the quantity she ate, she is going to experience rising AST and ALT. She's going to experience renal failure. She potentially could then develop coma, seizures, and death. And there are certain interventions one could make. Many of these are not well-described in the literature, but they involve intravenous milk thistle, high-dose penicillin, H2 blockers, sometimes liver transplant. So here's the interesting issue. If I am the only human on the planet who has the knowledge of 3,500 mushroom species and their clinical effects and treatment, that doesn't scale very well. Ideally, you'd want to turn that expertise into an algorithm so that an image can be submitted to a cloud-hosted service, and then a set of probabilities suggested. This looks like it could be an Amanita, and, oh, of patients like this, here's how they have done using various modalities of treatment to help, again, not replace the human, but augment human decision-making and democratize access to specialty knowledge. Let me give you some additional examples. Mayo Clinic has developed about 60 algorithms in the last two years. So in the fields of cardiology, we can diagnose ejection fraction, hypertrophic cardiomyopathy, pulmonary hypertension, even acute COVID-19 infection from a one-lead, six-lead, or 12-lead ECG using algorithms with an AUC of 0.9 or better. So what does this mean? And, in fact, literally, you could say, oh, I have picked up a device. I'm going to send that telemetry to a cloud-hosted service, and I'm going to be able to be told your heart pump is weak or strong within seconds, and then seek the appropriate care. Similarly, in the field of radiology, we've analyzed 200 million radiology images to develop the largest in history map of what is a normal distribution of fat, muscle, and bone in radiology images can be used for diagnosis. In the field of radiation oncology, we have taken the entire historical data set of all head and neck tumors treated at Mayo Clinic and developed an algorithm that creates an auto-contouring profile so that instead of 16 hours of human time, it takes about one hour of human time with an AI-augmented decision support system to create an auto-contouring profile, a linear accelerator program for delivery of radiation oncology treatment and complex head and neck tumors. In the field of oncology, we have a model with 84 input variables that predicts which women will develop breast cancer and identifies a subset of them for which therapy today can avoid breast cancer entirely. And of course, we know that these kinds of AI algorithms are gonna be used in a variety of novel virtual care settings, delivering not only standard, what I'll call ambulatory or outpatient care, but helping us with serious and complex, helping us deliver care at a distance in non-traditional locations, such as the home. The notion of this Mayo Clinic platform is to build those Lego blocks that I showed you so we can empower a whole variety of new investigations and new cures and reach more people than ever before using these algorithms and the various data flow components I've outlined. Here's one example of how we can do something entirely novel. Typically, serious and complex illness is treated in a bricks and mortar facility. I would argue this last two years of COVID has taught us that bricks and mortar in some circumstances can be replaced with clicks and orders. If one has remote patient monitoring, the right telemetry to understand a disease state and its progression, the right command center where you have the right experts, the right dashboards, the right care plans, the right supply chain and workflow processes, the right staffing, you're able to deliver a variety of novel therapies in non-traditional places. Mayo Clinic believes in starting small, thinking big and moving fast. We started with one patient and this particular patient in Florida had hyponatremia, serum sodium of 102, 87-year-old person. We proved that through this combination of remote patient monitoring and dashboards and algorithms and expertise, we could treat this patient with high acuity illness safely in the home. We then moved from one patient to 10 patients, 10 patients to 100, 100 to 1,000. As of today, we've treated 3,000 patients with serious and complex illness at a distance. And we've achieved the same outcomes, the same quality, the same safety, markedly reduced cost, markedly reduced readmission rate, markedly reduced nosocomial infections and other complications. And for example, in terms of complications, very common for the elderly to sundown. You move them into a non-familiar surrounding and they will get confused, fall, break a hip. If you're in the home, sundowning basically disappears. And so imagine this, as you think about the future of delivery of care for liver disease, what are those disease states you think you could care for with the right telemetry at a distance safely? I will tell you in General Mayo's experience is about 30% of patients who present for inpatient admission can be cared for in the home. It's not everybody. It's those disease states where the patient's sick, but is unlikely to really go bad in five minutes, right? If you need time to move them to a different setting of care, you can. So that means congestive heart failure, COPD, issues of say infectious disease like pneumonia, they work very well at the home. Somebody with a life-threatening ventricular tachycardia, you know, not so much. So ask yourself this question about hepatology, because imagine this future, as we've talked about, more algorithms, more data, and more non-traditional settings. Couple of things, just as we think of these new models, you do wanna make sure that as you're delivering this care, that you have the right people and the right supplies. Mayo Clinic has established a community paramedic training program to create a new cadre of caregivers who could come to the home, deal with some of the setup, the physicality of medicine, help with the supply chain. In the last two years of COVID, we've also seen regulatory reform waivers that have enabled us to deliver different kinds of care in different settings without necessarily the limitations of licensure of the people or places that we had in the past. And as I've also mentioned, there's the cultural changes that have occurred as well, where patients are now expecting care to be virtual in many circumstances because of its convenience and what it does for remarkable patient satisfaction. But let's talk for a moment about social determinants of health, ethics, and bias. One of the great joys about delivering high-acuity care at a distance is it requires us, in the case of this Advanced Care at Home program I described, to go into the home and understand a bit about who is in the home. What are the food sources? What is the cleanliness and safety of the home? Is that patient receiving all the appropriate family support or externalities around just the healthcare treatments that they need? Similarly, as we deploy algorithms, we need to take into account all of the varieties of patients we're seeing, including such things as socioeconomic status. What I'm showing you is a project that Mayo Clinic is just launching in collaboration with a number of other partners to create a nutrition label. This notion is, if you pick up a can of soup, you could say, I know the calories, the sodium and the fat in this soup. I can decide to eat it or not. If you pick up an algorithm that has been developed for hepatology, do you know how it was developed? Is it just a black box? Do you know if it's going to work for the patient in front of you? You need that level of transparency. So what we'll work on over the next several quarters is this idea of a label so that every algorithm comes with a set of well-defined performance characteristics and the ingredients that went into it to create it and to validate it so that you will be able to make a decision as to its appropriateness. Here's just a quick non-hepatology example. Recently, Mayo Clinic developed something called the Houses Index, where based on the address of the patient, we can actually understand for that neighborhood, what is the access to food? What about water and air and environmental issues? What about access to public transportation or specialty medical care or internet access? We recently used that index to stratify a large patient population into four quartiles of socioeconomic status. We then evaluated a very well-known algorithm. And what we found was when you look at the lowest quartile of socioeconomic status, the algorithm performs miserably. And very likely that's because the algorithm didn't take into account some of the stressors around a socioeconomic status that doesn't allow compliance or refills of medications or easy access to doctor's follow-up visits or telemedicine telecare. So I tell you that because as we think of this bold new future, we're all gathering more data and we're creating novel algorithms and new care delivery mechanisms. Let's meet the patients at their level of technological comfort and sophistication. Let's take into account social determinants of health and socioeconomic status. In doing so, we're going to achieve a very high quality of unbiased care and we are going to ensure the patients truly get the right care at the right time in the right setting. So I wanna thank you for listening today, but I also wanna make the following conclusion. I believe this future that we are in the next six quarters going to explore together will be increasingly digital first. It will not in any way diminish the importance of humans or human contact, but it will give our families the tools they need to get that care they want and need in a much easier fashion than in the past. And as I use that example of the mushroom ingestion I just had to work with, taking the world-class specialists and hepatology and making their knowledge available to all who need it in a democratized fashion through increasing use of data and virtual care is truly a bold future. And I think we would all agree we're all in this to make healthcare better globally. And these tools that I've described to you today are going to help us achieve that better future. I wish you a great rest of the conference.

American Association for the Study of Liver Diseases

AASLD

Liver disease prevention

Hepatology advancements

Artificial intelligence in healthcare

Mayo Clinic innovation

Digital healthcare solutions