Demel Ji Mehta pt 2

00:00:05 Speaker: I've started to do this a little bit with my patients, who are reasonably comfortable. But how to ask the AI question because I know they're going to go home and, talk about it and get on GPT or they're going to get on with their kids. I've even gotten to the point where I say when my note appears in the chart, just make sure you see the beginning of it, the diagnosis and, then put that in there. So at least you have a clear answer. John, I'm curious, have you done that yet? Have you, have you talked to patients about the proper use of AI? I have not actually. That's a really good point. You know, because when they come to me. we always talk about, what will be the diagnosis, why we do this, right? and then, you know, where will be the good resources for them to seek, a lot of times I say that, Nccn has the patient, website as well. So that's actually really reliable resources. so I think the AI actually patient education that can be really huge. I wish that there will be some sort of, you know, forum education on that. but I think that's a really good point. this idea of like triaging and, and, how do you, you get to this, the right information for us, for our trainees, what do you do when you have a fellow? And I don't know if you work with fellows, but obviously, you know, the the job of a fellow is getting difficult. And we're seeing burnout dramatically increase in, both residents, students, med students, residents, fellows. And, you know, there's been two replies to this. One is to say, buck up, we did it. You can do it. And I don't agree with that approach. I think the world's different today for them than it was for us and for prior generations. And I think the other is to say, well, what's the underpinning of this? And they I think part of it is they don't they feel like they can't it's too much to know. And they're overwhelmed and, what do you tell a trainee right now about how to kind of have a manageable practice of reading? Do you have any advice for the fellows on on information. Kurt, I'll start with you. I do. I mean, I think one a couple things that I, that I tell the fellows is that when they, this is their time to learn, take advantage of it. Take time to read, take time to read about the patients before you see them. And kind of like the stories that, they've had and the treatment that they've had so far. ask questions. this is the time to do it. the way that I learn and I just kind of this is my bias. I learned from my own cases and my own mistakes. So like, if they're doing like a certain, like block of like lung cancer, say, you know, you're working with someone that's only doing lung cancer, take advantage of this month to really learn, what are the right pathways and algorithms for a Pd-l1 positive non-small cell lung cancer, negative. Pd-l1, Pd-l1 negative small cell. really kind of like a use this month to really feel comfortable with some of like the elementary questions with non-small cell lung cancer. You're not going to come out of this month being an expert, but at least you might know the tools to kind of deal with very common frontline scenarios. The other thing that I tell them is start learning where you can go for information like this is the time to get those colleagues and those mentors that are going to be lifelong friends and colleagues that you can bounce questions off of. you know, when I'm rounding. And one of the first things that I kind of always tell the fellows is don't be scared to like, ask the experts in the field. I mean, they're very approachable. they're honored that you're asking them questions that maybe all of us are struggling with. I can tell you many examples of cases that I've seen both inpatient and outpatient where I don't know what's going on or I need help with the case. And I'm like, I go look up the research and say, this guy has published a serious amount of data on this topic. Talk to him or talk to her. You know, she's very approachable. And, I think you learn so much that way. Mhm. I also think, another, tidbit would be know, where to go to get information or to like synthesize it I really like research to practice, you know, love has a great engaging conversations and you can filter it by like lung cancer, breast cancer. So if you're doing that lung cancer block, listen to all, you know, all the recent, podcasts that he's delivered on lung cancer. That's something that I tell the fellows, I also tell the fellows to again, instead of reading the entire New England Journal of Medicine, there's like a, New England Journal of Medicine, section where like, kind of like synthesizes all of like the recent, take home messages from the, from the published literature, which I think is incredibly helpful as well. It kind of like gives you take home messages. I like that. I mean, I was looking this up there were, I think in, in two thousand, there were six hundred and fifty trials. And now there's, you know, three thousand oncology. And that's just trials. When you look at papers twenty some years ago, it was I think I saw sixty thousand papers, oncology, depending on how you search and. And last year it was three hundred and fifty thousand. then trying to sort through that and then see the apparently it was the most retractions of papers ever last year as well. And so looking at that data and then trying to keep up with something you read that may have been, you know, retracted gets gets complicated. I started telling our fellows, something I do, I have a forty five minute drive to and from work. And so I've started using the voice version of I use ChatGPT. just saying, hey, over the last month or last couple of weeks, give me the ten most relevant papers in Her2 positive breast cancer that and give me the citation. And then it talks to me while I'm driving. And then, when I get back, I have the transcript and I think some of the fellows have started started to do that. So trying to be efficient in your day has been, I think AI has helped us a lot from from that end. to kind of get to this point of like, what's enough this, this practical AI, in oncology, I think we talked a lot about this, that, oncologists I've sort of met, they're sort of on two fronts. I think everyone here is on the AI is going to help us front. And then there's some on the other side is the, you know, this hasn't been helpful. and I'm not so sure it will be. And there's too many hallucinations, and whatnot. How do you, I'm curious how you interact and what your interactions have been with colleagues. have you had situations with colleagues of different, whether it's different generations or different, backgrounds that, that are either pro or con, with AI and, and, And if so, why? Jan, maybe I'll ask you this. Yeah, I think we we talk about the AI, the good things and the bad things. first of all, I say that the knowledge and, I do think that sometimes I, think there is a point that we should say it's good enough, you know, and for, I think especially for physicians, because I think sometimes we feel guilty that we did not read every article published this week. We did not read, every email and send it to us from different, companies and podcast and all that. Right. but I think there is a point that we have to realize what will be the, the tears we need to learn from, you know, and the daily practice and, or use the, AI Because we're human beings, we have to realize that there is a limit. So I think from me, you know, I have been like, you use AI to, you know, for you to keep up with the breast cancer. And I think I have been using that one is that I'm using that to really update me as a standard care. So, you know, because the standard care, it's updated so frequently now. I think that's essential, right? So that's the I think that's a the base how we can practice, you know, as the best care that that, you know. So now I use Nccn, you know, I encourage my fellow, you know, and read Nccn every time when they have a, you know, a case, right. So just go back because, you know, sometimes even in a couple of weeks. And then the change, I also encourage them that especially during fellowships, since they have time, you know, actually you read the discussion part, you know, and I found that actually is quite helpful. You know, sometimes that may be a little bit old, but at least you can know like why we have this, you know, and practice pattern, right? So why, for example, you know, melanoma and we used to do axillary dissection. Now we don't do axillary dissection. You know, mostly people just do the sentinel lymph node biopsy. So why is that. Where is the trial. Right. Because the trial was done, you know, and ten years and or longer than that. So now we actually don't talk about that a lot of those publications anymore. And then two is that, you know, I think the AI is really useful and to keep us posted about the, the practice changing trials, like you mentioned that you use that to for the Her2. I think, if you, use the AI open evidence to find, you know, the, twenty twenty six what is the practice changing studies, you know, for general chemo actually that's quite useful. Right? So and we treat pancreatic cancer, colon cancer, lung cancer. Obviously you cannot you know, you cannot know every study, but at least you know what the practice changing study are. Over the past a year or two or even longer. So that's actually a little bit easier for us to catch. And then third, I think as we talk that it's collaboration. So I think a lot of times I feel like, you know, this is like Kerr mentioned, now we're feel very comfortable. the first line, second line, or we have our own cancer research center. You know, we can see whether patients are eligible for any clinical trial at our cancer research center. But after that, you know, there is a limit. So I think we need to reach out. You know, we need to reach out to people to see that whether there's any better care, people can offer to the patient, you know, whether the plan I generated is the good plan. So I think that in that case, I do think that AI has its use. But I think the use sometimes can probably is, you know, still limited. I think the collaboration, communication, human communication is still is, is still is a key. It kind of speaks to the issue of humility in, knowing what you don't know, and then reaching out and sharing that, you know, enough and then using that, that, you know, community, that we're lucky, we're lucky to have. I remember talking when I was a fellow calling up Larry Norton at Sloan Kettering thinking I just had access to and I, and I did, he answered. And like you said, Curt, he answered the question. And, and to this, you know, I think today could I just call. I probably could, which is, you know, we're lucky, lucky to have that, in these last few minutes, because it really seems like you're both so thoughtful and about this. I want to just we think about information and oncology that's already overwhelming. and, and we're getting kind of bombarded with, but I remember, I think, well, Abraham Varghese, who wrote that was written several books, but, you know, wrote that book about the Aids epidemic is my own country. And I heard him speak when I was a resident and he said, you know, good doctors reading their field, great doctors read outside their field. And I wonder, in this era of information where it seems like too much. How much do you use? Either the tools you have at your disposal for, you know, the, the, the regular websites to AI to read outside your field for our patients, because our patients do have all sorts of issues that involve the entire body. And you wonder sometimes, is it the cancer? Is it the chemo? Is it, you know, some weird toxicity or is it something else? I'll start with Kurt. Have you used, AI or any other tool that way? one thing I would say is that we're, we're at the infancy of AI. I think right now, like some of the topics that we've talked about, low hanging fruit, like dictation and voice recognition. That's low hanging fruit for AI. I think it's when we get to things like themes that you're talking about right now is like, what are the, how do comorbidities fit in? And how do we, not only know the oncology part of it, but tap into some of these other intricacies of the patient presentation. That's where AI is really going to be powerful. I can tell you like, I'm using it a lot in terms of just like, again, we're all in that same position of a patients coming to us with a host of issues. And, you know, the last time that I treated diabetes was a, you know, an internal medicine resident, right? So it's like, I don't even know some of the new meds that are out there and the interactions. So oftentimes with AI and again, I don't even know if I would call it AI. I'm trying to do like a, you know, a medication check with the chemotherapy that they're on and they're kind of like home medications for diabetes. I'm always doing a little bit of like a pharmacy check there. And I do use AI for that. I've been trying to use it for patient education material as well. And I'm going a little bit off topic here, but, one thing that I find, AI going to be very helpful. Hope hopefully we can use it for is just like, patient education and patient monitoring. I think it'll be, it'll be very helpful for, eproms and kind of like finding out like this patient based on like this wearable device that they're wearing or their HRV that's changing. You might want to call them into clinic for like a check. So that's where I envision kind of like AI going. And again, right now we're, we're having a little baby building blocks, but, very powerful, very powerful. I'm a proponent as the, is the take home message. the, the company that I spent half my time with reimagine care, we think a lot about e pros and devices and how do you manage, how can you start, democratizing access to therapies that we know are really effective. You know, these complex therapies, device specifics, even, you know, Car-T cell, how do you start doing these things outside of the big centers, into the community where seventy five percent of patients are treated. and you're right, using technology and using AI to be predictive analytics, to kind of put together all that data to safely manage and monitor patients is, is a great way to use it. I think my take home message is cautious optimism. I think we always have to be cautious with any new technology. AI being one of them. but yeah, enthusiastic. But just we have to be cautious, with the power of it. I think my take home message is that, I think as a human beings, we have nature to resist. You know, when AI starts, I know that myself and some of my colleagues feel like this is not good. but I feel like in the future there will be two, two types of oncologists, oncologists who knows how to use AI and oncology. I think, I believe that those people who knows how to use AI, you know, it's probably going to win. so better just start to learn how to use it. It's true. Right. Join or go home. I'll say my take home message is that, it's great to speak with, oncologists who are so thoughtful and experienced about, about all of this. And it kind of gives us hope that that the field is, is continuing to adapt really well to this information. I think we're doing the best we can. And, for our patients, for our trainees, for our colleagues. so thank you for joining me and for joining targeted oncology.