Eddie: [00:00:00] Hello and welcome. You are listening to the ATS Breathe Easy Podcast with me, your host, Dr. Eddie Qian, also the host of the ICN and Podcast. Each Tuesday, the ATS will welcome guests who will share the latest news in pulmonary critical care and sleep medicine.
Whether you're a patient, patient advocate, or healthcare professional, the ATS Breathe Easy podcast is for you. Joining me today is Dr. David Gazal and Dr. Shaan, who will discuss the ATS AI task Force in some recent these things coming up for the ATS conference for people to get involved. Dr. Gazal is a pediatric pulmonologist and a world renowned pediatric sleep expert.
Who is currently the dean of the Jones c Edwards School of Medicine at Marshall University and actually a past president of the American Thoracic Society. Dr. Shaan is a professor of pulmonary critical care and sleep medicine at the University of Miami, director of their MD PhD program, and the director of the Office of AI and Medical Education at the University of Miami.
In addition to chairing a number of committees [00:01:00] including including this task force. For the American Thoracic Society. So welcome to both of y'all. It seems like you guys have a lot of, a lot of experience and a lot of things to talk about in this space. Hi,
David: Thanks for having us.
Eddie: I, I think the, the thing for me when I was first, like looking at this and looking at the materials to prepare for this episode was, was just, I think just at first, a very general question. We're here to talk about the, the ATS AI task force. What, what is this task force and, and what really drove the ATS to try to prioritize having a voice in this space?
David: I'll, I'll tackle the beginning and then Shirin Dr. Will follow through with some more extensive details. But the important thing is we are a. Society that is meant to serve the membership in the best possible way through its three major domains, and of course the [00:02:00] community domain as the fourth domain.
So one is education, one is research, and one is clinical. And in all of those domains, we have seen extraordinary propulsion of AI taking huge steps into incorporating itself into the fabric of almost everything that we do. And it would be al we will be almost remiss as ATS, as a leading, as the world leading organization in respiratory medicine and the critical care medicine and enslaved medicine.
We would be remiss if we didn't have one, a better understanding of the needs in the membership. Regarding AI. And so how to best serve those. Two, how to best incorporate AI into the fabric of ATS by providing innovation and fostering collaboration and projects that may lead to novel discoveries and novel ways of educating or of clinical providing clinical service to our patients.
And three, how to also [00:03:00] incorporate AI into. Many of the activities that he has, for example, B Corp is one example. We are very proud of Corp and it's time of service promoting, epidemiology and training in epidemiology. But at the same time, we have all these wonderful projects that leads to consensus statements and guidelines and so on.
These are opportun, opportunistic, events that leads to the use of AI. So you can see where. AI will almost become an integral part certainly in the moving years. An integral part of everything that we do at ATS. So it, we need to mark a time that we get the first step into infra a into this area, and one, by doing this as a task force, we are doing two things.
One, we're listening to what the members want. 'cause that's the most important thing. That's why we have a society. And two, [00:04:00] we provide them an exposure as to what is currently in place. And by doing that, inform them of opportunities and initiatives that they could foster through their assemblies or through any other mechanism that the ATS provides.
Shirin, would you like to add on onto what I just said?
Shirin: Absolutely. And I think one of the roles of the task force is to also bring about a sense of coordination of all the activities that are going on. Many of our members already maybe engaged in using AI tools and in their clinical arenas and research and education.
And for ATS to have this role as a society to bring members together and to sort of coordinate across the various pillars of the platforms to ensure that we deliver like a consistent message, I think for, from our society and for our members and to provide them opportunities. To engage even more so in the society.
And I think the [00:05:00] task force role is to look across the mission statements of ATS and to really think through how AI tools can not only be incorporated into our society, into how we practice pulmonary and critical care and sleep, but also in some ways ATS has always served as a guide where members can go to for.
Questions around how do other people do things or what do we as a society think about doing something? And so eventually thinking through governance, thinking how these tools come into the health system. ATS may or may not wanna play an active role in it, but as a task force, it's important to think through these concepts along with expertise from our various members and reflect that.
To the board and then decide as a society where our positioning should be with respect to a very rapidly developing field. So there's no one person that can keep up with everything that's happening in the world of ai. And we really do need a society like we do need American Thoracic Society to [00:06:00] be one of the key leaders in this field.
Eddie: There's one thing I was gonna comment is it seems like this, this task force is very wide reaching and very daunting, so I'm glad we have such accomplished people leading this task force to kind of lead us, lead us into the future there on this podcast. Even before this, we, we've talked a lot about.
AI and some of the nuances and AI can include LLMs or large language models, NLP, natural Language processing. We've talked about machine learning models. Is there a, a focus or a line that you all are drawing as far as this AI task force? Or is this gonna be still just again, you need a whole committee?
'cause we're gonna cover everything.
Shirin: We need a whole committee. It's not that we're covering everything in the sense that we are not in the, we are not gonna be reinventing the wheel. We are not going to be directing tasks on, let's say, let's do this machine learning project, or, although somewhere down in the future it may be necessary to do clinical guidelines around machine learning.
And ATS does need to have [00:07:00] a say in that. But I think what is really important as a task force is to, to identify the needs. Of, of our membership and to see where ATS can use its collective might, if you will, to try and meet those needs. And that's where the task force becomes important. We do have a wide range within the task force.
We now have. Subcommittees a group of individuals who have expertise and interest in education, which would then be looking at these AI tools and thinking through implementation and perhaps guidelines and guidance from ATS about some of these tools. Educating the membership about the use of these tools in the realm of education, educating our fellows in the next generation.
There are people who clearly, as you said, are using it for machine learning purposes of, have been doing it for a long time. So while the public thinks of AI as. Just happening in 2022 when chat GPT came on the field. The reality is machine learning has been around for a very long time. And so that expertise [00:08:00] though while they will continue to do their research, maybe their role in ATS is going to be conveying how to evaluate and critically appraise machine learning products and predictive analytic tools.
As they become more and more used in our health system. So how can we evaluate the effectiveness of these products? And, and ATSI, in my opinion, should have a role in providing guidance to perhaps health systems that may not have access to all the information and all the expertise. So there's a lot that we can do as a society, and part of the role of this task force is to think through.
The gaps in knowledge or the gaps that are needed that we should be addressing at ATS and we can't address everything as you mentioned. There's just too much.
David: So another aspect that Dr. Chan that I would like to point out and maybe it reflects the very unique speed at which things are happening, is the fact that most patients today.
Consult AI before they come to a physician. [00:09:00] And we need to therefore not only to prepare the workforce the our communities of p pulmonologists intensivists and so on, and sleep physicians to, to those facts and to be ready on how to address those elements, but also to educate our patient, particularly our, our specific patients suffering from the diseases that we take care of.
Of how to best navigate through an environment that is all the time under modification is undergoing changes all the time. And that can be very disorienting for an individual that may not necessarily have the information at, at, at the tip of their hands or the tip of their computer to be able, or the experience to really critically assess the evidence as it is presented.
And that too becomes important. As it, as we interface with other partners whether it is other agencies, whether we interface with other [00:10:00] associations and collaborate around the world, we are. An international society that covers the world. And there are many of our fellow societies around the world with whom we have very significant contacts that may actually benefit from one of us, the ATS in particular, taking a lead on this area.
Start facilitating engagements with other societies to create a more homogeneous approach to the diseases that we take care of. Both in all the three domains that Dr. Sha and alluded to, but also in the interactions with patients in the more wide in, in a wider se setting.
Eddie: This, this is all very exciting and I know I, I do definitely do wanna get to some of the things that the task force already has planned in the upcoming conference and beyond.
But you, you know, I, I'd be quite remiss, you both of you all have such vast experience already in this field and, and maybe not speaking on behalf of the [00:11:00] task force, but maybe as individuals, what, what kind of. Questions or what kind of topics do you think are important for us to be addressing moving forward since you, as we've talked about, that this is such a broad field.
Shirin: That's a tough one. It's an easy one and a tough one. I think you have, you need more than an hour to talk about topics, but I can at least try and address it from my perspective. In AI and medical education, I, I primarily am engaged in undergraduate medical education, but also, of course, in GME and CME activities.
And the immediate need that comes to mind is as much as I like to believe that everyone is spending the entire existence. Thinking about AI topics and learning about AI tools, that is not the case. So the media need actually is in many ways training the workforce, but also training our faculty and people who interact with our students, whether at the undergraduate or the GME level.
Because many times our students are actually ahead of the [00:12:00] faculty in their use. Maybe not quite their they, they use it readily, but whether or not they use it with a critical thinking angle is, is to be debated at times. But that's okay because the role of all of us is to educate ourselves about these tools, when to use it, when not to use it, the ethical limitations, the regulatory limitations, and then.
Be a voice within our health systems because ultimately clinicians are working with these tools and they should be the ones directing its use and there should be the ones thinking through how this impacts the health system, how it impacts patients, how it improves or exacerbates. Disparities among patients depending on their access to these tools.
And so I think the more that we can educate and train the workforce, and the more that these national, international societies play a role in that for their membership the more powerful it's going to be. We are, [00:13:00] we as a, we can have a voice then as clinicians in determining the course of many of these tools.
I don't know about you, but every day. Without exaggerations. Every day there's an email, you know, we have developed XX platform and would you like to use it or would you like a demonstration of it? And, and you know, it, we think it's gonna be useful for you. And some of them are expensive, some of them are not.
But the reality is we need to be equipped to understand what these platforms are doing and how useful they are. They're not. So that's just a quick glimpse on the other hat that I wear away from ATS, but definitely relevant,
Eddie: I think, I mean, what you're saying about. You know, the trainees are sometimes the head of attendings.
I'm not, I'm not that old also that, not that far into my career. But even so now, the trainees are saying, oh, I, well I got I, I've never heard that before. And they'll say, oh, I got it from this x, y, and Z platform that I know is an AI platform. And yeah, I'm like, oh, that's, that's interesting. I always make a mental note to go and check that one out.
But there's always a new, every my list just [00:14:00] gets longer and longer it gets. It grows faster than I can actually, it grows too fast with these tools,
Shirin: but, but you know, the other problem is that we as human beings have a tendency towards, you know, what's called automation bias. As soon as it's there for us on the computer or written down, we believe that this may be the truth.
And the reality is, of course we know there's a lot of problems with these platforms. There's a lot of good things, but there are a lot of problems as well. And so being able to not. Let the machine do the thinking for you, but you do the thinking while using these machines appropriately, I think is very important.
Skillset.
Eddie: I completely
David: agree. Let, let me just add Addie maybe another aspect that I think is important, particularly in the context of a TA, the ATS being a scientific society, and that is the. Unique ability that we can use today. Tools that facilitate personalized medicine, more precision based both on the diagnostic level as well as the selection of targeted interventions.
[00:15:00] And so expose our membership. To exactly how these tools operate, their pros and cons, the caveats that are associated with this, the critical settings that we need to impose and filters that we need to impose not to become overly enthusiastic about some. By, but by the simple fact that understand what is behind the machine in order to understand how the machine operates.
All of these are very important components that will make us better by being able to understand how AI can be a wonderful adjunct, but it can be also a terrible enemy. And in that context particularly as it relates to research you know, we are talking today omics as if there was a very natural thing.
Well, multi omic pathways can be extraordinarily powerful and can define for a particular patient a specific genomic target or proteomic target or metabolomic target, and be able to [00:16:00] intervene and show that this works is fantastic. But it can be also failures. And how to navigate this and how this, the machine learning that has gone, or the neural networks that have gone, or the LLMs that have been IMP implemented in order to get to that result will actually dictate the quality of what we get and.
That understanding needs to be somehow explained, presented, and continuously improved upon as we move through the society. And we generate what I hope will be ex terabytes of data petabytes of data that are very important to to everybody, but at the same time have the limitations that I just mentioned.
And so that navigation. We as a task force and the ATS as the leading scientific society in our field needs to really show the way to everybody else and also make sure that we, in doing so, we do [00:17:00] not lead anybody astray.
Shirin: Yeah. Yeah. I'm gonna add one more topic. 'cause on the clinical domain, I think it really becomes important.
Many of us already are seeing it within our electronic health records that we have these clinical decision support systems that are multimodal AI, that are actually using these, variety of. Combination of natural language processing and machine learning, and even LLMs into already baked into the electronic health record.
And as busy clinicians even, even, even as busy researchers who may do clinical practice, there is very little time for people to sit down and think about whether or not this tool that has been deployed is in fact. Doing what it's supposed to or adding to your knowledge base or not. And I think it really is important.
A good example of that was sepsis models that were introduced into electronic, into, into a particular electronic health record in 2017 without actually testing. External to the electronic health record. In other words, [00:18:00] that kind of testing did not happen until 2021 for a particular sepsis model.
And lo and behold, the tool really was not that useful. Right? And so there's a lot of time and effort that goes into deploying these systems. So I think not that at S'S role in necessarily is to evaluate every single tool that's out there, but to really give its soci the society, the ability, and even the community.
To ask questions and check in with each other and have experiences and collaborate, hopefully in the future, collaborate on studies that validate some of these tools that are being deployed. I think that is a, is a nice place to be as a society.
Eddie: Yeah, no, absolutely. And I think both of you have highlighted a, a little bit of a.
I, I would say maybe of even a friction between like traditional medicine practice, which is slow, and then the ve an evolution of ai, which is a revolution that's really, really fast. And so sometimes these things butt heads. So I, I don't envy the position that you are [00:19:00] in, but I would look forward to what kind of solutions that you'll have for us for in the future or guidance that you'll have for me.
So but but I do wanna turn a little bit to the future and let's look at the near future in. I think maybe only a month or a little bit less than a month from when this episode releases. We'll be at the ATS International Conference, and my understanding is that, that you guys have things planned at the conference.
What, what if I, I'm going to the conference. What kind of things can I be looking forward to? I.
Shirin: And absolutely it's, and it's not just us two, it really is a very talented group of individuals and their names will appear on, on programming of course. And I don't want to mention every single one just in case I forget someone.
I don't wanna do that. But we have many of our junior and senior members joining into and volunteering their time. The big picture right now for the conferences that we have, what's called an ATS AI Innovation Lab that has a permanent spot [00:20:00] in the exhibit hall. It is a very nice central spot and all members can visit it during exhibit hall hours. In that space, we will have fun activities like trivia questions that ATS members can engage in and just to sort of check their knowledge about various ATS tools.
There will be presenters there as well from industry, not just. Talking about their product, but really talking about case use and hopefully evidence behind the use of their product. So that is definitely something that we are looking forward to. And then around that. Is workshops that have been developed by a lot of volunteers.
And with the workshops it's thematic as I mentioned before. So on Sunday the workshops will be around AI tools and in education, and we have about two hour session. There's a nice room that people who are. Sid can go into that room and we'll have demonstrations and [00:21:00] use cases where our presenters will be talking about how they use these AI tools in their day-to-day existence, especially around the topic of education.
On Monday, we're going to have the use of these AI tools in research. Again, we have a nice two hour workshop and people can go in and see how people use these tools for the purposes of brainstorming. Around a research project, doing the background literature search doing writing up specific aims, even analyzing data, using some of the existing tools without, of course going into detail statistics, but just showing how these AI tools can improve the workflow.
When you're thinking through the life cycle of a submitting a grant, for example, or analyzing data from your research. And then on Tuesday the workshop will be around the use of some of these AI tools in clinical medicine. So looking at some examples in radiology, which is obviously some tools [00:22:00] are being deployed to aid in diagnosing.
In both chest x-rays and CT scans, we will look at ambient AI tools and how clinicians have been using it, and the pros and cons of using these tools. And also these decision clinical decision support systems that exist and, and just talk about again. Benefits and limitations. So it is really jam packed.
most of these workshops are two hours. They're open to anyone first come, first serve. But we're also streaming it to the exhibit hall in our location that I mentioned, which is the AI ATS Innovation Lab. So even if you can't make it to the room, you can catch it and it will also be recorded and preserved for future use.
Eddie: it seems like there's a lot going on. Is there a, I think I know the answer to this, but I'll just make sure. Is there a particular like level of training that these workshops or tools or otherwise are targeted for? Is this really kind of like everybody would [00:23:00] benefit or what, what kind of, what kind of member, what kind of people are you looking to detract your
David: already?
As, as the beginning, as the beginning. Yeah.
This, the purpose is to really, this is the first little anchor that we're putting within ATS and we rushed to do it because it was so important. Things are moving so fast that we wanted at least to establish a little presence. This is really a little, little press.
It. We hope that this will fulfill three major expectations. One is that the ATS will start having a, an understanding and an awareness to AI across its all its platforms. The second is that any member. We'll be exposed one way or the other to the potential participation learning opportunities of ai and that we will see these areas thrive over the next few years because they, they're clearly exploding everywhere else.
So they're going to explode no matter what with us. So at least have them [00:24:00] coordinated and integrated within the society. And the third thing is to hope that assemblies will take on leadership. On their areas and start really developing the field so that they then are the content producers of the majority of the AI that the society can actually deal with.
And so we are, all we're doing is to putting a little bit of a enzyme and a little bit of substrate and hope that the rest, the conditions, the exhibitor, the. The enthusiasm of all the, the members coming to the conference and those that could also watch it either on a, a, asynchronously all of these will generate the kind of excitement and drive that will foster the development and propagation completely out of our control.
That's the time that the task force will have fulfilled its role and we'll rapidly wade into the, and disappear into the west and into the [00:25:00] sunset. So, so really the idea really is to minimize. To foster an initial surge in, in, get engaged and get engagement across all the, all the society without any preconceived assumption that anybody needs to know anything.
There's nothing better than a naive pair of eyes to look at things because we ca we have our biases and we carry our biases. And so. A new pair of eyes that has never touched on it and says, well, wait a second, guys. This is bogus, or This is wonderful and I love it because A, B, or C, which are the three things that nobody thought about.
So, so that's what we hope.
Eddie: Sounds like I'm the perfect, I'm the perfect person for this. 'cause everybody describes me as super naive. So sorry. Dr. Shaza, what you say?
Shirin: No, not at all. No, no. You're not super naive. No, but I think that the reality is that there probably is a little bit for everyone, even at the most [00:26:00] expert level.
I think if they wish to participate it's, it's gonna be a workshop and therefore at the end or even during, there's an opportunity for the person who has a certain expertise. To, even if they're not actively engaged in the initial content creation to bring their perspective in and talk about their experiences, I think we'll all benefit from that.
So but of course we, the assumption is that we're coming in all as beginners rather than aiming for, you know, super expertise level. There's also a lot of programming outside of this that's happening in ai. And if you actually just. Do a search on the program and just put in AI as the key word. You will see that there are symposiums and keynote speaks of speakers and meet the professor sessions that are planned with ATS and that, of course, is also going to be curated and, and shared during the AI innovation lab and in the exhibit hall so that people know.
Where to go if this is an area of interest for them. And the other thing that I encourage is it's your, we will [00:27:00] have surveys and it's your opportunity for the, the society membership to give us feedback, not just on the workshops, but on the needs. What is of interest? What do they want to learn? So that for future conferences, as David said, either the task force or hopefully.
The various assemblies take on that challenge and provide the content and programming that is needed by the membership.
Eddie: There, there, yes. It seems like there really is a lot going on in the conference, but Dr. Gza, you're, you are talking a little bit about beyond the conference and moving towards the future, so, so what kind of things or how can members ATS members, community members.
Get involved or what can we expect beyond the conference, beyond a few months?
David: Yeah, it's a great question. There's several ways of looking at this. The first one is obviously within the range of assemblies and activities within each assembly. [00:28:00] There's a variety of podcasts, of discussions, review of literature or, or publications and so forth.
So the assemblies can engage into. Deciding that they're going to take on AI as one of the topics that they would like to explore and do this. And you know, that that level of cooperation and fostering of a more continuous interaction beyond the conference is what we hope will take place. The second is that this will foster project.
Projects within the society by membership to propose unique solutions to specific issues that would benefit, like, for example, from ai at any level, education, research, or clinical. And that that can be implemented through the unique segues that are. Facilitated by the society. The third is that the society [00:29:00] itself, and its the leadership through the assemblies, the board and and beyond will establish a guidelines as to and a set of priorities for the society itself.
Ultimately without necessarily circumscribing them to a very rigid set of frame, but with enough guide as to what the society can provide. And differentiated from what memberships can do by seeking alternative routes to enrich the society or enrich their, their, their activities with, with ai.
I hope that this is you know, it, it's reminds me a little bit of when, and, and I will, that will tell you about my age. I was a physiologist and I started this integrative physiologist. And at one point in time, back in the late eighties, I was told that physiology was descriptive and that I needed to start looking at mechanisms and therefore I needed to do some degree of gene expression.
Today we laugh at this because a single gene [00:30:00] expression and some changes in northern or whatever we, we needed to do was not, clearly not mechanistic. But that was the, the impression. Then I moved forward 35, 40 years and we find that we have now a complex, very large complexity in the systems that we treat.
And AI is uniquely poised to help us decipher those components. And I hope that by bringing AI to the membership and by bringing AI to the A-T-S-A-T-S can actually. Help in the processing of solving unique problems that facilitate exactly the missions that the ATS is, is, is meant to to deliver.
So, so that's kind of what I'd like to say.
Eddie: Absolutely. And that those, these are, those are really good thoughts. I don't know. Dr. Chaan, do you have any other kind of closing thoughts on this or moving forward? Types of thoughts?
Shirin: No, well said Dr. Gelle. Thank you for that. I think really [00:31:00] the moving forward or closing thoughts is if there are members that are interested, I would really welcome them reaching out to myself or to David.
And, you know, while the members of the task force are sort of already determined because we have to have these subcommittees to make this happen it does not, there is plenty of work to be done and I think anyone who wants to be engaged or who, wants to present their or volunteer their time and their skillset to this endeavor.
More than welcome. So please reach out and we are happy to connect with you.
Eddie: Yeah. And, but even potentially we could leave some of the contact information in the show notes if we can get from those from you after the show. Yeah. But I, I think I would like to thank everybody for joining us on today's ATS breathing episode.
Please subscribe and share this episode with your colleagues. As we've already mentioned, the conference is right around the corner. Conference thoracic.org today to register. Members get a discount on their conference registration. Become a member. Renew your [00:32:00] membership to take advantage of the savings.
But we will see you next time. Thank you all.
Shirin: Thank you. Thank you.