Latest / The SHEA Podcast / AI in Antimicrobial Stewardship

Transcript

AI in Antimicrobial Stewardship · The SHEA Podcast

• 0:08
  SPEAKER_00: Welcome to another episode of the Shade Podcast.
• 0:11
  SPEAKER_00: My name is Sheena Kandayan, and I'm the director of Antimicrobial Stewardship Program at Grady Hospital in Atlanta and an associate professor of medicine in the Division of Infectious Diseases at Emory University.
• 0:21
  SPEAKER_00: I'm also a member of the Chade Antimicrobial Stewardship Committee, and I'll be serving as today's host and moderator.
• 0:27
  SPEAKER_00: I'm pleased today to talk with Dr.
• 0:29
  SPEAKER_00: Richard Medford, East Carolina University Health's first chief medical informatics and digital health officer.
• 0:34
  SPEAKER_00: Dr.
• 0:35
  SPEAKER_00: Medford is a renowned infectious disease specialist with over two decades of experience in clinical practice and research.
• 0:41
  SPEAKER_00: Dr.
• 0:42
  SPEAKER_00: Medford has been at the forefront of integrating advanced technologies into healthcare to improve patient outcomes and combat antibiotic resistance.
• 0:50
  SPEAKER_00: Dr.
• 0:51
  SPEAKER_00: Medford's innovative approach and dedication to improving healthcare have made him a respected leader in his field, continuously pushing the boundaries of what's possible through the intersection of medicine and technology.
• 1:02
  SPEAKER_00: Welcome, Dr.
• 1:03
  SPEAKER_00: Medford.
• 1:03
  SPEAKER_00: Thanks for joining us today as we talk about what artificial intelligence means for antimicrobial stewardship.
• 1:08
  SPEAKER_00: I'm excited to hear your opinions on this important topic today.
• 1:13
  SPEAKER_01: Thank you for that warm welcome and introduction.
• 1:15
  SPEAKER_01: And I'm certainly excited to talk about this field that's really special to me.
• 1:20
  SPEAKER_00: Great.
• 1:20
  SPEAKER_00: We're going to go ahead and get started with some questions.
• 1:24
  SPEAKER_00: Can you explain how AI is currently being used in the field of antibiotic stewardship and what specific problems it aims to solve?
• 1:32
  SPEAKER_01: Yeah, absolutely.
• 1:34
  SPEAKER_01: I think before we even talk about the specific applications of AI, I think it's really important to give us a better understanding of what we mean when we say artificial intelligence.
• 1:45
  SPEAKER_01: You know, it's synonymous oftentimes, I think, in the outside world to mean you may hear different terms like machine learning or deep learning or computer vision, natural language processing.
• 1:57
  SPEAKER_01: All of these are just terms that fundamentally fall under this umbrella of artificial intelligence.
• 2:03
  SPEAKER_01: And I think in our field, it's not as important as it is in like computer science to distinguish those things.
• 2:09
  SPEAKER_01: So practically, the way I look at it, look at it is through the lens of am I trying to predict something?
• 2:15
  SPEAKER_01: So I have some data now and I'm trying to predict something in the future.
• 2:19
  SPEAKER_01: Am I trying to generate new data that doesn't exist and trying to create new tasks through that possibility?
• 2:25
  SPEAKER_01: Or similarly, am I using something like ambient technology or ambient AI, which really is just listening to me?
• 2:31
  SPEAKER_01: So when you think of your Amazon Alexa or anything like that or your Siri.
• 2:35
  SPEAKER_01: So those are kind of the three main buckets in my mind.
• 2:39
  SPEAKER_01: So when I think about how it's been applied to ID and more specifically antimicrobial stewardship, we've really focused on, I think, predicting antimicrobial resistance and really focusing in on that prediction bucket.
• 2:54
  SPEAKER_01: So things like optimal antibiotic selection, really identifying which patients are at risk of X disease or X condition, and almost for like real-time monitoring of interventions.
• 3:05
  SPEAKER_01: So on a very high level, we really haven't touched the other buckets in the field.
• 3:11
  SPEAKER_01: And I think that's where, you know, eventually we need to head towards.
• 3:15
  SPEAKER_00: So that's really interesting.
• 3:17
  SPEAKER_00: I think I'd love to hear a little bit more about the prediction piece of AI.
• 3:22
  SPEAKER_00: But one of I think our biggest questions in antimicrobial stewardship is how to change behavior of um of physicians around antibiotic prescribing.
• 3:32
  SPEAKER_00: What are your thoughts about using AI to develop some of these interventions and whether or not you think it could actually predict providers' change in behavior?
• 3:41
  SPEAKER_01: Yeah, that's a really complicated question.
• 3:43
  SPEAKER_01: Um so I'll say definitely the prediction piece, oftentimes, as you alluded to, is the easier part, right?
• 3:52
  SPEAKER_01: It's really not that difficult nowadays to take data and run it through some machine learning algorithm or some model and come up with a really great model, meaning, you know, within the 99%, whatever it may be, I can confidently tell you that your patient's going to be septic or your patient's going to have a MERSA infection or something like that.
• 4:13
  SPEAKER_01: That's easy, uh, to be quite frank, as long as you have the data.
• 4:17
  SPEAKER_01: The difficult part really is the downstream kind of impact of that.
• 4:21
  SPEAKER_01: So, A, how do you appropriately fit that into a clinician's workflow so that they can leverage the tool?
• 4:30
  SPEAKER_01: B, what type of trust do they have in terms of how that algorithm or that model was developed in the first place?
• 4:35
  SPEAKER_01: So is there kind of transparency with respect to the model itself?
• 4:40
  SPEAKER_01: And then C, how comfortable are they with the technology at all?
• 4:44
  SPEAKER_01: Right.
• 4:44
  SPEAKER_01: So those three facets definitely play a huge role, not only in stewardship, but any type of, I think, technology implementation.
• 4:51
  SPEAKER_01: And it takes time.
• 4:52
  SPEAKER_01: It's change management, it's the change of culture, it's really um getting people more familiar with the technology, educating folks on what it means and what it doesn't mean.
• 5:02
  SPEAKER_01: And I think like, you know, things like this are definitely the first step in doing that, but we definitely need to push that out further in the clinical setting to be able to get what I call machine learning to the bedside, really make these algorithms translate into something meaningful for our patients.
• 5:17
  SPEAKER_00: That's great.
• 5:18
  SPEAKER_00: I think I I agree.
• 5:19
  SPEAKER_00: I think it's a huge challenge to do that, especially with the various uptakes of technology.
• 5:24
  SPEAKER_00: Some people are just more comfortable with technology that with technology than others.
• 5:28
  SPEAKER_00: And I think it's a it's a real challenge to try to sort of even just something as simple as an order set, you know, is hard to implement in an institution.
• 5:35
  SPEAKER_00: So, you know, as we go into more complex technology, I I can imagine it being a little bit more of a challenge as the more complex it becomes.
• 5:43
  SPEAKER_00: So, what role do you think that data analytics plays in enhancing stewardship programs?
• 5:48
  SPEAKER_00: And how does AI help in making those more informed and effective decisions?
• 5:53
  SPEAKER_01: Yeah, so that's also I think a nebulous term, right?
• 5:56
  SPEAKER_01: Analytics.
• 5:56
  SPEAKER_01: What does that mean to you?
• 5:58
  SPEAKER_01: What does that mean to someone else?
• 5:59
  SPEAKER_01: The way I think of it in my mind is it's kind of threefold.
• 6:03
  SPEAKER_01: It's the way we collect and manage the data first, then how do we actually analyze the data, and then how do we visualize that?
• 6:11
  SPEAKER_01: So those are the three main steps in my mind that allow someone to leverage data analytics.
• 6:16
  SPEAKER_01: So oftentimes we'll have a really great data set, and then so what we're really good at is describing what I call descriptive analytics.
• 6:25
  SPEAKER_01: So what we can do is describe that there's an association or there's a certain amount of, let's say, an infection or a hospital-acquired infection on a unit as an example.
• 6:36
  SPEAKER_01: The next step in that kind of path, though, is not describing it, but really predicting is that infection going to happen, right?
• 6:45
  SPEAKER_01: And so there's this thing called the analytics ascendancy model, moving from what we call descriptive analytics to predictive analytics, and then finally prescriptive analytics.
• 6:57
  SPEAKER_01: So we've moved from describing the situation to then predicting it, which is kind of where the machine learning and the AI comes into place.
• 7:04
  SPEAKER_01: I can now say with some confidence that yes, not only is there an infection in that unit, but it's going to happen on that unit because of XYZ.
• 7:12
  SPEAKER_01: And then prescriptive analytics is really what we what we talk about when we're discussing, well, how do I actually prevent it?
• 7:19
  SPEAKER_01: So it's kind of this holy grail of analytics.
• 7:23
  SPEAKER_01: And that's really where I think AI can really help us.
• 7:26
  SPEAKER_01: And that's where we need to be moving.
• 7:28
  SPEAKER_01: So getting out of the descriptive part and really moving to that predictive and prescriptive part.
• 7:33
  SPEAKER_01: And in my mind, out of all of that, the most important part is the visualization part, actually.
• 7:38
  SPEAKER_01: So all of that can be important.
• 7:40
  SPEAKER_01: But if I can't show it to my clinician or I can't explain why this really works really well and give it into give it to them in a digestible manner, then it's never gonna kind of move forward beyond that great model that you've created.
• 7:54
  SPEAKER_00: So can you maybe um give me an example of that in antibiotic stewardship?
• 7:59
  SPEAKER_00: Is there something that you're doing in your institution that you can maybe share with us?
• 8:03
  SPEAKER_01: So, yeah, I think a really great example is some of the work that we've been doing around antimicrobial resistance and identifying what I'll call hot spots uh in the community.
• 8:13
  SPEAKER_01: So if you take that from the lens of a descriptive analytics, we can say that patient X from certain zip code has MRSA in their community or has CRE in the community.
• 8:28
  SPEAKER_01: But really, we're not saying that patient is at risk of it.
• 8:31
  SPEAKER_01: We're just saying there's hot spots in the area.
• 8:33
  SPEAKER_01: So if you take that work, and this is kind of extrapolated from COVID, right?
• 8:37
  SPEAKER_01: So, like if there's COVID in one community and it's it very prevalent in that zip code, maybe we should start vaccinating that community, just as an example.
• 8:46
  SPEAKER_01: But that's descriptive.
• 8:48
  SPEAKER_01: Let's move this to now the predictive part.
• 8:50
  SPEAKER_01: So it's not only what environmental factors that you have at risk.
• 8:54
  SPEAKER_01: So your community has a lot of MRSA or your community has a lot of CRE, but then there's these individual level predictions that we need to make.
• 9:01
  SPEAKER_01: So I am from that community, I am from that zip code, but I also have diabetes, I also have hypertension, I also have a BKA, et cetera.
• 9:10
  SPEAKER_01: So that combination of taking that descriptive analytics and the other kind of personalized factors and combining the two, that allows us to now make a personalized prediction or personalized stewardship or personalized medicine, whatever you want to call it.
• 9:23
  SPEAKER_01: And at that level, now I've moved from the community down to the patient.
• 9:28
  SPEAKER_01: And that's supposed to help us inform now, you know, should I really be treating them with a carbapenum or can I get away with, you know, ceftraxin or something like that?
• 9:37
  SPEAKER_01: So that's kind of how I move from descriptive to predictive and then hopefully prescriptive in the future.
• 9:43
  SPEAKER_00: Yeah, so that's almost uh operationalizing the definitions of hospital acquired or community acquired type infection, but it brings in more of the personal attributes of a patient or comorbidities.
• 9:54
  SPEAKER_00: So I think that's going to be a very valuable thing in determining what antibiotic to use.
• 10:00
  SPEAKER_00: That's I think that's a great way to use it.
• 10:02
  SPEAKER_00: I'd love to see that more in the future.
• 10:05
  SPEAKER_00: How do you think so far the implementation of AI and antimicrobial stewardship has impacted patient outcomes?
• 10:11
  SPEAKER_00: And do you are you familiar with any success stories or notable improvements in patient care?
• 10:17
  SPEAKER_01: So I think you alluded to this earlier, but really um the area of AI and antimicrobial stewardship, I think, is nascent.
• 10:26
  SPEAKER_01: There's a lot of opportunity for us to get more involved in this space and really start, I'll say the implementation phase.
• 10:35
  SPEAKER_01: So while a lot of people are really good at creating models, they stay at that phase, right?
• 10:40
  SPEAKER_01: So, um, or they stay in the research phase.
• 10:43
  SPEAKER_01: So we've published this really great paper.
• 10:45
  SPEAKER_01: It shows this really great AUC, and uh we know it can effectively change whatever outcome you're looking for.
• 10:51
  SPEAKER_01: Where we struggle a lot, actually, is then moving that into the clinical site or into the clinical space.
• 10:58
  SPEAKER_01: To be fair, I don't know of many in the US, to be quite frank.
• 11:02
  SPEAKER_01: Where I've seen some success actually is in Asia, and uh a lot of I think published papers showing implementations of stewardship outcomes related to kind of IV to PO switches and um appropriate targeting of antimicrobial therapy coming out of South Korea, China, Taiwan.
• 11:22
  SPEAKER_01: So it seems like there's a huge emphasis over there, and I think that's what we need to do over here now.
• 11:28
  SPEAKER_00: And just to follow up on kind of moving that forward from the stage, what kind of training do you think is indicated?
• 11:35
  SPEAKER_00: What kind of where what kind of resources exist for clinicians or antimicrobial stewardship programs to access to actually try to try to move into that implementation phase?
• 11:46
  SPEAKER_00: Or are there any?
• 11:48
  SPEAKER_01: Yeah, that's a really that's a really great question and hotly debated in terms of like what resources are required, right?
• 11:55
  SPEAKER_01: I can share with you, like I was having a conversation earlier, there were a group of us from all over the country advocating that we need informatics embedded into ID and antimicrobial stewardship.
• 12:06
  SPEAKER_01: So we need more of our ID docs and more of our ID clinicians and providers and pharmacists to understand this field and really get a better sense of how they can contribute as opposed to someone pushing it on them.
• 12:19
  SPEAKER_01: And I think that's the huge barrier here.
• 12:22
  SPEAKER_01: I don't think everyone needs to be an informaticist or informatician, but I definitely think as we move the needle and technology becomes more pervasive, that we have to be more educated as a kind of a group and as a society in the space.
• 12:35
  SPEAKER_01: So there's like practical things about implementation that I think we just we lack and we need better education on, we need more training on without actually going the full length of actually getting a fellowship or anything like that.
• 12:49
  SPEAKER_00: Do you think that there's some responsibility on the part of the electronic medical records to do some of that?
• 12:55
  SPEAKER_00: Or do you think it really lies squarely on the clinician?
• 12:59
  SPEAKER_01: I think it's twofold.
• 13:00
  SPEAKER_01: You're absolutely right.
• 13:01
  SPEAKER_01: So definitely it lies on the clinician, but I definitely think the EHR vendors have some responsibility in terms of sharing best practices, sharing how it could be successful at one health system and therefore translating that to another so that both health systems don't have to recreate the same thing.
• 13:20
  SPEAKER_01: So that's where I think they have to have accountability.
• 13:23
  SPEAKER_01: But at the same time, I think if we don't even have a voice in it, then no one's actually listening to us in the first place.
• 13:30
  SPEAKER_01: So really we need to have that voice to say that's what needs to happen in order for these EHR vendors to listen to us.
• 13:37
  SPEAKER_00: Yeah, that's great.
• 13:38
  SPEAKER_00: Thank you.
• 13:38
  SPEAKER_00: So what kind of what are the some of the challenges and limitations that you face when integrating AI into antimicrobial stewardship practice?
• 13:45
  SPEAKER_00: Again, recognizing that it's early, you know, I think that addressing that may you may have answered that question through training, but any other thoughts as how these can be addressed as well?
• 13:55
  SPEAKER_01: Yeah, we touched on it.
• 13:57
  SPEAKER_01: I think um, like if I had to highlight the integration with clinical workflow, that doesn't require anyone to be an informaticist.
• 14:05
  SPEAKER_01: That really just requires a clinician to understand how they do their job and ensuring that you're partnering with your IT person that's helping you implement it so that they can understand your workflow, so that they can embed it in the right spot at the right time, so that it's the most effective for you.
• 14:22
  SPEAKER_01: So that's a huge thing, I think, that's overlooked oftentimes.
• 14:26
  SPEAKER_01: And like you said, not only for AI, but just any type of technology or order set or anything it may be.
• 14:32
  SPEAKER_01: Uh, we touched on this a little earlier as well, the interpretability and trust of the models as well.
• 14:37
  SPEAKER_01: I think that's a huge barrier.
• 14:39
  SPEAKER_01: So when you think of machine learning, and I'm going to focus on deep learning, which is a subset of machine learning, there is rarely interpretability in that.
• 14:48
  SPEAKER_01: But that is probably the most sophisticated way that we can create a model, meaning it will give you the best results.
• 14:55
  SPEAKER_01: But as clinicians, we rely so much on figuring out what went into that, right?
• 14:59
  SPEAKER_01: What were the cofactors that went into that in order for us to make that prediction?
• 15:05
  SPEAKER_01: So there's this balance and trade-off that we're going to have to make where either we'll become comfortable with not knowing, or we're going to have to stay not knowing and therefore use the models that probably aren't as sophisticated and therefore probably won't give as good of a result as if we had switched.
• 15:20
  SPEAKER_01: And that trade-off is, I think, something everyone struggles uh with.
• 15:24
  SPEAKER_01: But I think you'd be surprised to know that that trade-off happens in like everyday life almost all the time and we're not aware of it.
• 15:32
  SPEAKER_01: You know, I get that it's patients' lives, but it's not not everything is like life or death.
• 15:37
  SPEAKER_01: So I think we just have to get more comfortable with that type of nuance as well.
• 15:41
  SPEAKER_00: So looking ahead, what advancements do you foresee in the intersection of AI and antibiotic stewardship?
• 15:47
  SPEAKER_00: And how do you think it's gonna shape the future of healthcare and antibiotic resistance?
• 15:52
  SPEAKER_00: That's a loaded question.
• 15:56
  SPEAKER_01: Yeah, so primarily we focused on I'll say the prediction part, but if uh if you remember from the beginning, we were talking about the generative part as well as the ambient part.
• 16:07
  SPEAKER_01: I'm gonna hone in on the generative AI piece.
• 16:10
  SPEAKER_01: This is like an opportunity that I've never seen before.
• 16:14
  SPEAKER_01: I have not seen technology move this quickly, get embedded so quickly into clinical the clinical space.
• 16:21
  SPEAKER_01: And there's a massive opportunity for ID for stewardship to be able to leverage that type of technology.
• 16:28
  SPEAKER_01: And just for those who aren't familiar, generative AI really means like I am generating something new that did not exist.
• 16:35
  SPEAKER_01: And that could be text, that could be music, that could be um images, any anything you want.
• 16:41
  SPEAKER_01: So something did not exist, and now I am creating something new.
• 16:45
  SPEAKER_01: And specifically in the stewardship space, where I see it really helping is identifying through combing through like the massive amounts of charts without having a human needing to do that.
• 16:56
  SPEAKER_01: And what I mean by that is traditionally we've had to look at discrete data, meaning I need to enter this in a flow sheet row or something like that for me to understand if this patient was on an IV medication or a PO medication or what their time course was on their antibiotics, or what the clinician was thinking, and therefore helping inform what decisions were being done.
• 17:17
  SPEAKER_01: All of that is generally removed with generative AI and these large language models.
• 17:22
  SPEAKER_01: So we're able now to extract that data, present it into a meaningful fashion, either through note summarization, chart summarization, so that cognitively it's much easier for the for a person to do their job.
• 17:35
  SPEAKER_01: So I could totally see that making someone more efficient, right?
• 17:39
  SPEAKER_01: So they can focus on other things and perhaps spend less time doing some of the more menial or mundane tasks associated with stewardship.
• 17:47
  SPEAKER_01: The other, I think, large area where I see this playing a role is really aggregating massive amounts of data and having a way to comb through that that wasn't traditionally possible.
• 18:01
  SPEAKER_01: And it gets back to kind of what I said, but if you need all the microdata, if you need your radiology data, if you need your path data, you need all of that stuff just in one place to help you make a better decision.
• 18:13
  SPEAKER_01: I think I really think that's where the generative AI space needs to go and needs to be.
• 18:18
  SPEAKER_01: So I really see see that playing a huge role in the future in stewardship.
• 18:22
  SPEAKER_01: And then if I focus on the ambient stuff, and what I mean by that is there are a few large players in the market.
• 18:28
  SPEAKER_01: Most people think of ambient AI as really helping folks with clinical documentation.
• 18:33
  SPEAKER_01: So similarly, I think it offloads the clinician or it gets better data for you.
• 18:40
  SPEAKER_01: So it's able to say, well, you didn't really ask about X, Y, or Z that could help inform this decision that we need to make.
• 18:49
  SPEAKER_01: Therefore, you probably should ask it.
• 18:51
  SPEAKER_01: And then that will come in your note.
• 18:53
  SPEAKER_01: And also, it's not a note that you're typing, it's automatically generated for you.
• 18:57
  SPEAKER_01: So I think the combination of those two technologies then help better inform your predictions and vice versa.
• 19:04
  SPEAKER_01: So they really work hand in hand.
• 19:05
  SPEAKER_00: I think I I just as I listen to your talk, I'm just waiting for the time that I could go to the my electronic medical record and and uh say, hey, Epic, can you tell me what antibiotic this patient's been in on for the last month and what the days of therapy are?
• 19:19
  SPEAKER_00: I think that's good.
• 19:20
  SPEAKER_00: That that will be a game changer for a lot of our analytics as you as you had described and would accelerate the implementation pieces of intervention.
• 19:27
  SPEAKER_00: So I'm really excited for that to happen in the in the future for sure.
• 19:32
  SPEAKER_00: As of right now, what are your favorite AI tools?
• 19:36
  SPEAKER_01: Well, I'll give you some personal ones.
• 19:37
  SPEAKER_01: How about that?
• 19:38
  SPEAKER_01: So there's things that make my life a lot easier on a day-to-day basis, and I'll I'll share share some anecdotes, I think.
• 19:45
  SPEAKER_01: So ChatGPT, which I think everyone is probably well aware of.
• 19:50
  SPEAKER_01: Um, that's probably my favorite tool and my go-to for anything that I need to do in my day-to-day job.
• 19:56
  SPEAKER_01: But I use it for like interesting things.
• 19:59
  SPEAKER_01: So let's say I have some leftovers in my fridge and I really don't know what to make with it.
• 20:04
  SPEAKER_01: So I will take a picture, I'll put it into GPT, and it will say, create a meal for me.
• 20:10
  SPEAKER_01: It'll create a recipe for me, and then we'll cook it.
• 20:14
  SPEAKER_01: And that that'll be our dinner.
• 20:15
  SPEAKER_01: Now, if you think of that more practically in the clinical sense, um, the way I my colleagues use it in like our lifestyle medicine clinic, for example, they'll say, Well, I have this diabetic and they have heart failure.
• 20:26
  SPEAKER_01: Create like five different meals that are these different themes for these patients that allow them to still enjoy the things that they like, but are low sodium or, you know, et cetera, et cetera.
• 20:37
  SPEAKER_01: So practically, I think that's a really good one.
• 20:40
  SPEAKER_01: Um, and you can kind of have some fun with it.
• 20:42
  SPEAKER_01: The other kind of big one I use is uh Microsoft Copilot, but I use Microsoft Copilot in my like office suite of tools.
• 20:53
  SPEAKER_01: So PowerPoint, Microsoft Word, Microsoft Excel.
• 20:57
  SPEAKER_01: It actually generates my slides for me, it generates my um text for me within within the documents, it analyzes things in spreadsheets for me if I need it, and I just kind of ask it to do it.
• 21:09
  SPEAKER_01: Um, so that's been a game changer.
• 21:11
  SPEAKER_01: That one costs a little bit.
• 21:13
  SPEAKER_01: So your your uh organization has to be willing to pay for that one.
• 21:16
  SPEAKER_01: But uh, I think that's another good one that's really helped me in my productivity, to be honest.
• 21:21
  SPEAKER_01: And I think my final one is probably DOLI three um or Dolly.
• 21:26
  SPEAKER_01: That's also another generative kind of AI thing, but it generates images.
• 21:30
  SPEAKER_01: So what I like to do is I don't know, if if I have a really complex topic and I'm trying to share that with someone and I don't want to think about how to visualize that, I will just throw some words or the thoughts or whatever is being conveyed in there, and then it'll generate the image for me.
• 21:49
  SPEAKER_01: And then I'll use that kind of a teaching point or the springboard for further discussion.
• 21:55
  SPEAKER_01: So, yeah, those are kind of some of my favorite tools.
• 21:58
  SPEAKER_00: Wow, I I bet you save yourself a lot of time.
• 22:01
  SPEAKER_00: You're super productive and super efficient while I spend days on end trying to create presentations and images and God knows what.
• 22:09
  SPEAKER_00: Um, so that is, I think we have a lot to learn from you.
• 22:12
  SPEAKER_00: So, you know, I'm looking forward to a private lesson on how to use these tools.
• 22:18
  SPEAKER_00: But for now, I think that ends our podcast for today.
• 22:20
  SPEAKER_00: Thank you so much, Dr.
• 22:21
  SPEAKER_00: Medford, for sharing your perspectives and experiences.
• 22:25
  SPEAKER_00: It's a super interesting and relevant topic today.
• 22:29
  SPEAKER_00: And I think as infectious disease doctors, you know, I think that the given our volume of patients, the more efficient we can be to improve our quality of life is is best.
• 22:39
  SPEAKER_00: And I think that that you've given us a lot of uh uh you know information and tools that we can use.
• 22:44
  SPEAKER_00: So thank you so much for joining us today.
• 22:46
  SPEAKER_01: Well, thank you for having me.
• 22:47
  SPEAKER_01: Um thank you for allowing me to share my perspective.
• 22:49
  SPEAKER_01: And yeah, I'll just echo that.
• 22:51
  SPEAKER_01: I think there's so much we can do uh in the field, and we really need to take advantage of kind of this point in time to really leverage the technology to its best ability.
• 23:00
  SPEAKER_00: That concludes this episode of the Shea Podcast.
• 23:02
  SPEAKER_00: Thank you for tuning in, and we encourage you to check back for more episodes launching soon.