Can AI Improve Veterinary Diagnostics and Academic Efficiency w/ Candice Chu, DVM, PhD, DACVP, Texas A&M University

In this podcast session, I had the pleasure of sitting down with Candice Chu, DVM, PhD, DACVP, to discuss how artificial intelligence (AI) is transforming veterinary diagnostics and education. Candice, a clinical pathologist and educator at Texas A&M University, has been leading the way in integrating AI tools like ChatGPT into clinical pathology, helping to streamline workflows and enhance academic efficiency.

• AI in Veterinary Diagnostics: Candice explains how AI is being applied to veterinary diagnostics to improve accuracy and efficiency. She shares her experience with a custom GPT model she’s developed specifically for clinical pathology and how it’s making a real difference in diagnostics.
• Boosting Academic Efficiency: We also explored how AI tools are changing the way veterinary students and professionals engage with academic content. From simplifying research to creating more effective learning methods, Candice offers practical insights into how AI is improving academic workflows.
• Ethical Considerations: We touched on the ethical challenges that come with AI, including the importance of ensuring diagnostic accuracy, avoiding over-reliance on technology, and maintaining professional responsibility in clinical settings.
• The Future of AI in Veterinary Medicine: Candice shared her vision for the future of AI in veterinary practice and education, showing how it can reduce time-consuming tasks and help professionals focus on improving patient care and outcomes.

This conversation offers valuable insights on how to use AI to improve both diagnostic accuracy and educational efficiency. It’s a great chance to learn how AI can help you stay ahead in the rapidly evolving field of veterinary medicine.

Be sure to catch this episode and see how AI is shaping the future of our field!

• About Dr. Candice Chu (Including her social media and achievements)
• Candice’s Paper
• Undermind AI

Be Part of the Pathology Evolution: Stay informed on the latest in digital pathology innovations. Subscribe for more insights, become a member of the Digital Pathology Club, and get your complimentary copy of “Digital Pathology 101“. Embark on your path to discovery and progress in the fascinating world of pathology.

Aleks: [00:00:00] How do you see this changing how students learn?

Candice: Personally, I’m very optimistic on that. This would definitely cause a paradigm shift in how we learn and how we teach students, but it’s going to a positive direction that really boosts our efficiency. I don’t see why there’s a trend of skepticism of people shouldn’t be using AI, AI is everywhere.

I’m tired of it because you can take the airplane to UK. Why are you selling the boat to UK? I feel like there’s no excuse to doing things in a less efficient way.

Aleksandra: Welcome Digital Pathology Trailblazers. Today, my guest is Dr. Candice Chu, author of the recent paper of ChatGPT in Veterinary Medicine. So I saw this paper on LinkedIn and I’m going to give you a little bit more detail how we know each other from social media, spoiler alert, but I saw this paper [00:01:00] and I’m like, Oh, I need to invite her because she needs to tell me and you.  How she did it. 

Candice is a veterinary clinical pathologist and clinician scientist at Texas A&M University. and an online educator. So online education is where I met her for the first time on Instagram. And so I’m going to link to her Instagram platform. It’s a vet clean path professor. It used to be Vet Clin Path resident, but since I met her, she became a professor.  So that’s where a lot of her audience is. 

Welcome Candice to the show, to the digital pathology podcast. How are you today?

Candice: I’m doing good. Thanks for having me here.

Aleks: Thanks for joining. So let’s start with you. We always start with the guest.

Aleksandra: Welcome Digital Pathology Trailblazers. Today, my guest is Dr. Candice Chu, author of the recent paper of ChatGPT in Veterinary Medicine. So I saw this paper on LinkedIn and I’m going to give you a little bit more detail how we know each other from social media, spoiler alert, but I saw this paper [00:01:00] and I’m like, Oh, I need to invite her because she needs to tell me and you.  How she did it. 

Candice is a veterinary clinical pathologist and clinician scientist at Texas A&M University. and an online educator. So online education is where I met her for the first time on Instagram. And so I’m going to link to her Instagram platform. It’s a vet clean path professor. It used to be Vet Clin Path resident, but since I met her, she became a professor.  So that’s where a lot of her audience is. 

Welcome Candice to the show, to the digital pathology podcast. How are you today?

Candice: I’m doing good. Thanks for having me here.

Aleks: Thanks for joining. So let’s start with you. We always start with the guest.

So let the digital pathology trailblazers know about you, your background, and also your journey, because your journey to where you are right now is geographically and in [00:02:00] general, not that standard.

Candice: Yeah, sure. So I obtained my DVM degree in Taiwan at National Taiwan University and I came to the United States about exactly 10 years ago, starting from Texas A&M University that I pursued my PhD degree in veterinary pathobiology. And then after I finished the PhD, I stayed on for three more years to complete the veterinary clinical pathology residency.

I got boarded, and then I moved to Pennsylvania as an assistant professor of clinical pathology, worked there for about two years, and recently decided to move back to Texas A&M because I love Texas so much.

Aleks: Is it like, Weather wise, is it similar to Taiwan, where you come from? 

Candice: Yeah, look, weather wise, it’s pretty similar.  Taiwan I would say it’s pretty much as hot as in Texas, but Taiwan is more humid, so sometimes in the dry season. 

Aleks: Ah, okay. 

Candice: Texas, it feels better. [00:03:00] 

Aleks: Okay. So it’s not even that bad.

Candice: Yeah.

Aleks: So we actually met in person at the ACVP conference, American College of Veterinary Pathology Conference last year, and I was giving a presentation about AI in veterinary pathology.

I’m an atomic pathologist, so my. AA presentation was more like anatomic pathology and image analysis heavy, but I was comparing and contrasting the reach of digital pathology, static images in literature with what’s happening in social media and Candice, your profile that I featured was one of the most like visited the highest amount of followers on your profile.

And it was so cool because then I saw you. In the, in front of the microphone. And he says, I’m the owner of this account. I’m like, so great to see you in person. And then we took a selfie. [00:04:00] So that’s a little side story. But so being in this online teaching space, I think we are more exposed or like tend to use different AI tools more than people who are not so active in the online space.

So naturally, we start thinking, how can we leverage it for our science?

And this is the topic of today’s episode because you wrote a paper and the protagonist of this paper is your customized GPT for clinical pathology. Let’s talk about this paper. Let’s talk about what inspired you to write it. And then we’re going to dive into how you built this GPT.

Candice: Sure. So the first idea came from my career transition period. I left UPenn and in the process of [00:05:00]  moving to Texas and I figured well I have some time, and I always consider myself as a tech savvy person, like I’m the so called IT in my lab, like anything related to computer, I will jump in and solve the problems.

Since I have some time, and then AI is such a trending topic, so I spend some time reading papers about AI. And realize there are actually many publications for example, like review papers in human medicine side, talking about how AI especially generated AI, like chat GPT can be applied to… 

Aleks: I love this part of your paper where you say like, how many hours are being wasted in filling the electronical health records, which then they like print on paper anyway. But I love this initial part, I’m going to Google right now how many hours it was, unless you remember, do you remember?

Candice: I remember it’s 35 percent of their [00:06:00] time is spent. 

Aleks: Yes. Can you imagine 35 percent of doctors time is they either typing before, like during the visit or after? So they only work actually 65%. Of the time. 

Candice: Yes. Yes. 

Aleks: Pardon my math if I didn’t calculate it correctly, but I think I did. 

Candice: Yes. 

Aleks: Yes. I love that. Also a comment to a tech savvy and IT person. This is, if you want to do digital pathology nowadays, this is what you need to have, or you need to have a person like that in your lab because there is going to be troubleshooting.

Like as much as I love the technology, there is still a lot of troubleshooting. 

Candice: Yes. 

Aleks: But, of course, I interrupted you going back to my question, the paper. One more question in between, how much time did you have between the jobs?

Candice: Roughly remember, I think like several months [00:07:00] I have, cause it hasn’t been a good chunk of time since I started my PhD, like one thing after another, even after I took the board exam.

Or even after I finished the PhD program, I think I only had about one week of vacation before starting my residency. And then right after the residency, it was like board exam, finding jobs, moving across different states. So there was no time for me to really spend time on things that I want to learn more in the past 10 years.

So that was a good time for me to be like, okay. Now I have some me time. I can rethink about what will be my future goal of my career and what I want to spend time on to learn as a new stuff.

Aleks: So in your free time you decide to write the paper?

Candice: Yes. And I enjoy the process [00:08:00] very much. 

Aleks: That’s great.

So Did you like write the paper first or did you create the GPT first?

Candice: That is a good question. I think… 

Aleks: How was the combination of these two?

Candice: Yeah, I think I might have been doing both at the same time. And actually, it all started from that I agreed to ACVIM, the American College of Veterinary and General Medicine, to do a talk on AI in their meeting.

Aleks: You just recently gave this talk, right? I saw that in LinkedIn…

Candice: Yes. So I agreed to that. Yeah. And then in the process of preparing for this talk, then I realized, yeah, what I’m doing right now is actually the same thing as writing a review paper. Why don’t I just write a review paper, right? [00:09:00] Since it’s like the same thing. So I started to work on the review paper and somehow got the idea of oh, it would be cool to build a custom GPT because I read that somewhere in my literature search.

So I’m like doing both at the same time. 

Aleks: Okay.

That GPT, let’s talk about it. So you basically this is a custom GPT built on the open AI framework. So this is something that’s out there that people can already use. And it’s Clin Path specific, veterinary clin path specific. So walk me through the process.

How you did it, what tools did you need for it? Like everything Oh, first you need to have the paid account for GPT. Then you need to whatever tell me the process.

Candice: Yeah. In order to use the GPT, [00:10:00] you don’t have to pay OpenAI, but you do need to have account with them. But if you want to create your custom GPT, you need to pay them.

So that’s about $20 per month. And once you have that, you can create this GPT for your own use to set it to be a private account, or you can publish it as what I did for the Vet Clin Path GPT. And the advantage of using a custom GPT instead of just using the chat GPT is that you can provide a specific material.

To the chat bot so that they will only look for answers in the material you provided and you can even restrict like I only want you to look at information on these specific websites so that you can ensure the information that you’re getting your users are getting are accurate and it’s scientifically sound instead of just a random blog article [00:11:00] on internet and talking about clinical pathology,

Aleks: I think it touches on the main fear that people have using this for medical purposes. And I think many of the I call me included the people who are using this are like commercial tool users are afraid of the hallucinations. This is like the buzzword.

That, Oh, I don’t know if it’s going to start hallucinating about the things that I don’t know about, and I will not be able to verify. And I think it’s not advertised enough or not communicated enough that, Hey, you can prevent it by what you did, you can prevent it by telling it to just use the specific resources that I provided that you can tell it to reference things.

Where, like exactly on which page or in which paragraph it found the information? [00:12:00] So which material did you train it on? And because it’s okay, there’s this general large language model, the GPT, the chat GPT. Now we have the 4.0 or I don’t know if we have some turbo or something like the next level. So basically you leverage the power of this, plus you give additional data that this model is referencing, right?

Candice: Yes. So I use open source available textbooks that I found online that they are allowed, they will allow because if you look into the the Common Creation License, that they have different levels and you need to make sure they do have this, they don’t have this like non derivative level restriction because it’s in the gray zone. Like some people consider if you feed this information in AI [00:13:00] and I spilled out the answer for you, that’s like a derivative of the original work. And in that case, that will violate the common creation license. So I chose the material that specified that they allow this type of usage and also finding resources from the e-Clin Path website, because they also have this open CC creation not CC creation creative commons license as well.

Aleks: Creative commons. 

Candice: Yeah.

Aleks: Which by the way, I know that your posts on your Instagram are under one of the levels of the creative commons license. 

Candice: Yes. Yes. 

Aleks: Thank you so much for making it available for people. So if you guys want to cite a post from Instagram from Dr. Candice, Chu,  it’s possible. I think it’s the attribution or which one do you have?

Candice: I don’t remember what I have. I have to check it on Instagram right now. 

Aleks: We’re going to Google it right now.  Yeah. Basically it says [00:14:00] exactly what you can do and what you cannot. Definitely need to cite where you took it from. Which is what I did in my presentation. Cause I cited your Instagram. So this is really cool.

Did you have a lot of material to train the chat on? Because you were restricted by the license and by the availability of the material. How did you curate it? Where did you take it from? And what was it? Was it just articles? Was it books? Other than the website that you mentioned as well? What did you put in there and how much data was there?

Candice: So there are several textbooks. There are more basic ones like histology textbooks, but there are also a book that’s specified in clinical laboratory works. So that’s very relevant to vet Clin Path. And of course, there are e-com path websites as well. And how I train [00:15:00] this model is I tell this model that please provide those sources with their answer, sometimes chai GPT, they won’t like always listen to your instructions.

But every time if you ask a question, you got to answer, you can always ask the GPT can you provide the source? And then usually it will provide you if it’s from e-ClinPath, it could provide you a URL. If it’s from book, it’ll provide you the name of the book. And even, you can even ask which page is this information from?

And then it will tell you the answer. So you can always go back to your original source when you’re looking for answers.

And the other thing I tried to emphasize in that tool is this is a tool to facilitate the learning for veterinary students specifically for veterinary clinical pathology.

It is not a tool for your, medical consult to answer [00:16:00] questions for your patients or provide any medical advice.

Aleks: So how can students use it? What is, what would be the best use case to leverage this tool when you’re learning clinical pathology? Would it be more for veterinary students or already for residents or for both? How do you envision this? as an education tool for them?

Candice: Yeah, I think it’s for both. It’s the most obvious use user scenario would be if you attend the lecture of veterinary clinical medicine, And then medical pathology and then you have questions, but maybe you’re to shy. You don’t want to ask the instructor or you come up this question, you got home. You can always ask the question. In ChatGPT, because what it offers is offered by [00:17:00] that chatbot function is, it’s not like when you Google for answer, you can only put in keywords, one keyword space, the second keywords, and maybe you add some logical relation like, and or, but you always have to like parse through all the answers and find the right one for you.

However, in using this chat bot function is like you are talking to a. actual professor on your computer about your question, and you know that those answers were only from the textbook. So you can be 100 percent sure that this is the answer you want. So that’s the most obvious user scenario.

But there are also much more than that you can ask this chatbot to create some questions for you. Can you create some true or false question? Can you create some multiple choice question? To help me to learn, for example, acid-based interpretation in this case, and then more [00:18:00] advanced than that would be if you have a case, Let’s say you have a blood work and that could be like a JPEG or a PDF file.

You can upload those files to the chatbot and ask what’s your interpretation? So for this function, a lot of veterinary educators would be terrified. Because that means that is the end of all this take home homework open ended questions, and even for trainers, for graduate students there will be no more of those preliminary exams that you take home and go through five open ended questions and think about it for two days.

Now they can get the answer in five minutes.

Aleks: Yes, so how do you see this changing the education? By the way, a curiosity in Poland, I’m from Poland, and recently I learned that the [00:19:00]  government mandated that there will be no homework for kids. 

Candice: Oh, really? 

Aleks: I don’t know. Yes, and I’m like on one hand, I’m like, okay, will they then learn at school what they’re supposed to learn? And on the other hand, I’m like, yeah, there’s already eight hours at school and then they have to do four hours of homework. So I’m in between. I don’t know if it’s good or bad. I guess in 10 years we’re going to see what happened to the kids, but that’s like an official country mandate that there is supposed to be no homework.

Candice: Wow. 

Aleks: I don’t know where it’s gonna lead us. But definitely, yeah, it’s not something like, also with internet, that was already okay. You have questions, you can find the answer in the internet. But then, you actually had to rewrite it. Now we don’t even have to do that. You can upload a transcript of, This is my last paper! Based on my last paper, please provide another [00:20:00] paper that answers these questions. How are you going to, because you’re going to be teaching students, are you going to be using it as a teaching tool? How are you planning to design your curriculum? Disintegrate it, or in general, even if you’re CLINPATH, and GPT was not available, these tools are available, and people can customize or not even customize, like, how do you see this changing how students learn?

Candice: Yeah, I got a question from my talk at ACVIMS as well, that there are because I talk about how to use digital and AI tools to streamline your academic writing process from, identifying the paper preparing for literature review, and then actually writing it down. There are questions from the audience that they worry about oh, if all these AI tools, they can summarize the paper and generate the notes for you.

Would that [00:21:00] take away the critical thinking of our trainees? And personally, I’m very optimistic on that because I feel like what AI is doing is just saving your time to do those like redundant, repetitive works. But that means it frees you from those works. And now you have more time to actually look at the, for example, the summarize or the outline and to think about what’s the main question that is being asked in this paper.

And then you have more time to do your critical thinking. And one thing also proposed from the audience is that they now have changed their questions a little bit. So they don’t ask those obvious questions that you can use AI to find the answer. But they may present something and then ask you like, Why do you think the author will put it this way?

So more like,  maybe presented and [00:22:00] summarize summary made by the AI and then ask follow up questions from the students. So I think this would definitely cause a paradigm shift. In how we learn and how we teach students, but it’s going to a positive direction that really boost our efficiency.

Aleks: I think so too. I’m looking for a new AI literature search tool that I recently learned about. I will find it before the end of the episode, but so see, I’m thinking totally opposite. It’s gonna, the critical thinking and reviewing of these things. It’s going to be required, it’s going to be the thing that like is going to be now trained because all the memorization and retrieval of information is taken care of.

And then you go through it. And at [00:23:00]  the end, I think it’s going to be also with the use of these tools.

And we’re going to get to the ethical use, which is another point that you mentioned in paper. But I think there is going to be for ethical use. I see something that’s going to be super necessary is the emphasis of being responsible for the ultimate output of this, regardless what tools you used.

If you sign your name, be it on a paper, on a presentation, lecture, you’re giving, or a report in my case, it is your name and you are responsible. If you like don’t review what the tool provides. Put out it’s on you. It doesn’t matter what you use.

And I see it very much in the peer review process or even like scientific review process of my reports.

I write them. I think I did the best job possible. And then the second pair of eyes, my reviewer looks at it and finds like [00:24:00] obvious things. Which are range from like everything, right? Some redundant sentences, spelling mistakes, which, oh my goodness, if I didn’t have a spell checker as a non-native English speaker probably, I probably make still mistakes on social media.

I hope it’s just non-native speaker. Please forgive. There are now are the tools and It’s a reason and let’s say, let’s take this spelling. The reason I make spelling mistakes is that English is not my native language. But it’s also an excuse because there are tools that can help me overcome this characteristic of myself.

Let’s call it that way, right? And it’s going to be the same for other things, right? If there is a shortcoming, whatever. [00:25:00] It might be, lack of material experience spelling. There is going to be a tool that you can leverage to overcome this short. So there’s not going to be any more excuses.

I think it’s going to be I see it as leverage.

I see it as a tool that can leverage our, the stuff that we are great at like analyzing critical thinking, maybe then discussing and analyzing on the fly with the context that we have, like you mentioned, the description of a case and on the fly asking people to analyze, right? You can get maybe five minutes and Google slash ChatGPT, whatever you want, but then you have to defend your answer.

Candice: Yes, like it’s as a clinical veterinarian, you might have a challenging case [00:26:00] or you’re trying to recall something from your memory. And typically you do is you go back to, you leave the exam room and go back to your office and then you start to flip through the textbook or whatever. Now it’s the same of using that custom GPT, like you have those textbooks ready, but now you have a smarter way to search to find the information that you want. So I think by taking that approach, whether you use other people the custom GPT built by other people, or you have the ability to build your own GPT based on the textbook that you commonly referenced to those are just more better, more efficient way for you to find the information you need in your shorter time. So from that perspective, I don’t see why there’s a train of like skepticism of like :0[0:27:00] “Oh, people shouldn’t be using AI. AI is everywhere. I’m tired of it.” Cause for me, it’s you can take the airplane to UK why are you selling the boat to UK?

Aleks: Yes, yeah, that’s the audience, or, yeah you know. 

Candice: I feel like there’s no excuse to doing things in a less efficient way. In this AI era.

Aleks: Very much. Same like with a car. Okay. You’re gonna walk there and you’re gonna be late, or you’re gonna take a bike and you’re gonna be late if you can take a car.

Candice: Yeah. 

Aleks: Or like you mentioned a plane or anything.

And by the way, I found a website. Let me share it here. Oh great. It’s called, I actually learned about, I dunno if you already know about it. It’s called Undermined.Ai. And you need an account, you can have a free one. And it searches, [00:28:00] so let’s see if we can search for Candice’s paper.

“What is the use of AI in veterinary medicine? Can you restrict your answers to the literature from the last three years?” – Let’s see if it shows up your paper. Has some, okay. Okay. I don’t know if it shows. Yeah, but it’s just showing the application of AI in veterinary medicine has significantly grown. 

Candice: It’s asking you whether you’re looking for a specific type of publication.

Aleks: Oh yes. yes. 

Candice: It could review articles.

Aleks: Okay, yes, please include all reviews. So see, it’s even talking to me. It’s asking me I want to have specific specific [00:29:00] references. All reviews from the past three years. Oh, wow. It even tells me what I’m supposed to ask. So I have an account already, but I have not played with it yet.

Let’s see. I want to see if it finds Candice’s paper. But while this is searching let’s talk about ethics.

What are your thoughts on ethics and AI use? Because I think that’s another fear. First one is hallucinations. We kind addressed that. Is it ethical? It’s going to take away jobs. And in general, another aspect of ethics with, okay, what data was used and you were very strict in your selection criteria, used things under a certain type of creative common license.

Yeah. What are your thoughts, ethics, and Generative language, AI.

Candice: Yeah. One very popular use of AI tools in veterinary clinic is the note taking tools like it can record [00:30:00] your conversation with the client. 

Aleks: I love it. 

Candice: And then it will output in a specific format. A lot of people like to use soap note to take the progress note of the patient.

One obvious ethical problem is have you obtained. the consent from the client for the recording. Are they aware that you’re using this tool to record their conversations and you always have to, be the one to put the signature on it. Like you mentioned, like you will be the author of the work.

So you take the 100, you take the full responsibility. So you always have to review those notes after it’s been generated and then put your name on it. I think that would be like one ethical concern of using those A.I. tool. And the other thing we need to pay attention about is there are tools to provide diagnosis. [00:31:00]

For example, there are so many tools right now. It would interpret the X-ray. For you, and those are like veterinary radiology, AI tools that you can use. However, if you go to their website, you will notice that some of them are backed up by scientific publications. Some of them are not. So how would you know the interpretation and diagnosis they provide is accurate enough?

That would be another concern.

Aleks: Yes. So see, that is Yeah, that’s something that you need to do your due diligence on the tool, right? In case of your GPT, okay, there is a list of literature that this was based on. And by the way, we have our results. Yes. Let’s see what we have in a second.

But yeah, you have a list of literature. [00:32:00] And with image analysis-based tools, it’s a little bit different because what are you going to check? What images it was created? It’s a little bit, so it’s a little bit less. verifiable by a non-expert, whereas words, anybody can verify words and anybody can see, okay, this was based on these and these publications, the impact factor of the journal was this and that.

And these were the authors, right? Does it mean that the information there is 100 percent accurate? If you have read any like publications on how is the scientific peer review, then that’s another discussion for a whole other. A series of podcasts, but in general, this is like the best way we have to trust information.

There is a peer review process and, at some point you stop doubting and decide, okay, this is what I go with.

So yeah, let’s talk about the AI in podcasting. Academic writing and I saw on social media that [00:33:00] you got a chance to have a spontaneous presentation. Your main presentation was on the general AI topic and then you jumped in into the spontaneous presentation about leveraging AI tools for scientific, scientific work, and scientific literature creation or writing, how do you approach that? What tools do you use? If somebody is writing a review, for example, right now, and what tools would you recommend? And how would you approach the whole process?

Candice: Yeah, so the process, I divide that talk into several different actions, sections. So it starts with how you find those literature that you’re interested in.

So instead of using keywords to search periodically, I would leverage the RSS function so you can subscribe to the newest updated paper. For example, for me I have a section called AI in veterinary medicine.[00:34:00] So based on the people I provided those new articles will RSS platform. So I can just open a single web page… 

Aleks: So how do you do that? Where can you subscribe for RSS? Because I do it like a subscription on PubMed. Is that the same? And I have AI and digital pathology.

Candice: Yes, that will be the same. You can use the advanced search function and then create RSS based on that or you can just click on the RSS button of a specific journal that you’re interested in.

For example, like general veterinary medicine. they provide that as well. So that would be the first step. And then the second step is how you collect all the information and put into your citation management tool. I used to use EndNote and that is a proprietary paid software usually when you are associated with a university.

So you have access [00:35:00] to… 

Aleks: Institution, you can…. 

Candice: Yes. Yes. 

Aleks: I switched to Zotero when I left the university.

Candice: Yeah, me too. Yes. And what’s really fascinating about Zotero is it’s now integrated with so many other either digital or AI tools that are super useful in academic writing so that you don’t have to, find the papers or drag the paper from your folder into those platforms.

You just have to log into your Zotero account and everything in your library will be automatically imported into the tool that you want to use.

Aleks: So do you then use the AI things from Zotero or do you use separate AI tools?

Candice: Yeah, those would be separate your paper, separate tools. 

Aleks: It’s your paper. Yes.

Candice: Oh, really? That’s helpful.

Aleks: I found your paper. We’re gonna go back to Zotero, but we have to look what our Undermined found. So we know we’re super specific [00:36:00] and we have topic match. There is a score for topic match, and. And you are 85%, which I don’t know what the score is, but I’m going to figure it out by the time I make a review about it.

But there is Candice Chu, ChatGPT in veterinary medicine, a practical guidance of generative artificial intelligence in clinics, education, research 2024 frontiers in veterinary science, which is like super new. When was it published? Because the moment you published it on LinkedIn, it was still under review, right?

You had the author’s copy to share. And now it’s actually officially published. Let’s click it. Let’s show it.

Candice: Yeah, it’s actually published one day. Before my talk at CVIM, so that was very . 

Aleks: Oh, and congratulations. 436 downloads already. And I was one of the people who downloaded it. {[00:27:00] 

Candice: Thank you.

Aleks: 1,557 Total views. They took, accept the cookies. That’s okay. So yeah. Oh, and I love this one. There is like a special table of key takeaways. 

Candice: Yes. 

Aleks: If somebody is too lazy to read the paper, or if somebody already read it and forgot, like me, you can just go and take the key takeaways. There is a full table.

And I also love your graphic abstract. Your graphic abstract is so cute. It is the theme of the graphics that you have on Instagram. Actually, I recognize these little, the style from your Instagram account. And on the graphic for those who are not watching YouTube we have a division of Do you remember those circles where the biggest circle [00:38:00] is AI then within AI there is machine learning and then within is generative AI And then within our large language models and then there are those going to different uses of ChatGPT in medicine in veterinary medicine which obviously overlaps with uses in life And But yeah, so I’m going to be exploring, I actually have to write a paper that is overdue for a year already, don’t want to omit it, so I’m going to be talking about what paper that is, but I’ll try to use this Undermine tool and let’s see if I can somehow integrate it with Zotero or use it in parallel with Zotero.

I’m going to link to this in the show notes as well. Amazing.

So are you planning to have a module in your, what are you going to be teaching at Texas A& M?

Candice: So I’ll be teaching the second year veterinary clinical pathology [00:39:00] course. And because I just recently started my own lab, I am actually protected from teaching for the first two long semesters.

However, I volunteered to do a two hour lecture on electrolytes and acid base. Because that’s one of the topics that I’m interested in veterinary clinical pathology. And other opportunities that came up after I presented this work at the ACVIM is I was invited to lecture to all the new coming graduate students at the College of Veterinary Medicine at Texas A&M University about how to use these AI and digital tools.

To streamline their literature review and also academic writing process. And I think this would be super helpful for incoming graduate students so that they learn the most efficient way [00:40:00] at the beginning of their graduate study.

Aleks: Very much. I remember me trying to, so I started my PhD and residency after working three years in the clinic.  And I suffered trying to go through scientific literature, like it took me a long time before I was able to just extract the information I wanted and not be intimidated by the way the papers are written because they’re written differently. For a normal person speaking normal English, they’re written in an intimidating way.

Very scientific and specific language. And my mission is also like to present the information from science in a clear way so that, people can take action on it. Yes. And if the, there is a barrier I don’t want to say it’s a barrier because this is the way science is communicated in the science circles.

[00:41:00] But I think there is a missing piece of the chain to later take this form of communication and take it further to people who are not yet that advanced in their scientific career, be it by AI tools, be it, by what I do or what you do on social media to bring this information. To people who communicate in a different way than the language of scientific papers.

So thank you so much for joining me. Thank you so much for creating the Clin Path ChatGPT am I butchering the name? Vet Clin Path GPT? Yeah, vet ClinPath GPT. I’m gonna link to the vet clip of GPT in the show notes. So anybody who wants to see actually, who would be your target audience as well.

The anatomic veterinary anatomic pathologist who had to study clean path for the boards and then immediately after the boards totally forgot . And here, [00:42:00] truly yours is raising her. ’cause I remember when I was learning it for my boards, I was like, this is logical. This is like analytical. And then two months after, I’m like, what?

Acid based . How about I ask my clinical pathologist, colleague, colleagues, thank you so much for joining me. And thank you so much for doing what you do. Your students or future students and colleagues are up for a treat and keep trailblazing Candice. Thank you so much for joining me.

Candice: Thank you.

Aleks: And I hope to see you at some conference in person that we’ll have to do a live, live podcast. Where are you going next?

Candice: So I’ll actually be attending ACVP meeting this year. Oh, yeah. And I will be talking about the topic of how to use digital AI tools to help your academic writing. [00:43:00] In the works.

Aleks: Ongoing yet, but if anything changes, then this is going to be the place where you have a wonderful day.

Candice: Thank you!

Aleks: Thank you so much for listening. If you’ll stay till the end, you are a true digital pathology trailblazer and you do want to leverage that. AI for your work however it’s possible. I would encourage you to check obviously the paper and everything is linked in the description, but also check the literature research tool Undermined.AI and check it out. And let’s see if it’s helpful for you in your literature research and in your scientific paper writing. 

And I talk to you in the next episode.