Hey, guys. Welcome to the podcast. I have a very special guest. My partner in crime, Katrina Pellet, is here. She's MSL excellence lead for our company, MSL Mastery. And we're really excited because we have a new training program for AI that's amazing. And this episode is sponsored by MSL Mastery, so you have to go and check us out. Go to mslmastery.com where we have training programs for MSLs, and we have coaching programs and online courses, and it's just amazing amazing.
We're so proud of the progress that's being made. And today, we're gonna do something a little different. So I just got back from a conference, which was AI in medical affairs, and so many people have asked me about it that I wanna tell you about it. And who better than to ask me and be the host of this episode than the one and only, Katrina Pellet? This is a hostile takeover, guys, of the MSL Talk podcast. We're changing it to Patrina Talk, and Yep.
I'm really excited to introduce my guest today, Tommy Caravella. Yeah. Just attended this conference. Are you ready to get started, Tom? Oh, yeah. I I so many people have asked me, like, they couldn't make it. How was it? Was it good? Can you share it? And I'm like, you know what? I think we just should have an episode where we talk about AI and talk about some of the stuff that was covered, in the conference because I think it's really beneficial.
And let's plan on doing another one next year and the following year as AI progresses. That'll be such a cool follow on. Yeah. Oh, no. Totally. And here's the crazy thing. This was the 4th annual AI. That's so crazy. I didn't I didn't even hear of this before. Yeah. Me neither. But they they got their they they did a great job. So They're ahead of their time. For sure. For sure. Alright, Tom. So set the stage for us. Where were you at? What was it like? How many people were there?
Were there vendors everywhere? Just set the scene for us. Yeah. Well, it was a great venue. It was in Philadelphia. It was cold, but it wasn't in a hotel. It was in a place called Syto, SytoPhyl, SytoPHL, which is almost it was specifically like a place that you rent to do small meetings, conferences, seminars, not big. There was probably a 100 people there. There were vendors, really smart people.
I mean, it was just a actually a really intimate atmosphere that and and the venue, I thought, was really cool. It I was worried that I'm like, well, it's not a comp it's not a a hotel, so what's parking gonna be like? Parking right there in the building. It was really convenient and awesome. Okay. Awesome. So it was a cold day in Philadelphia at a cool, hip, modern location. Totally. So tell us a little bit about what was covered in the conference and which AI topics were discussed.
Yeah. So the like, I feel like they touched upon and covered hit on most of, if not all, of the major areas of AI. So for example, like, core capabilities of AI, use cases, a lot of talk about use cases, which I think a lot of people wanted to see. How are other organizations utilizing MSL I mean, utilizing AI for their MSLs, for their medical affairs department, for patients. Like, it was like, really touched on all the different areas that and use cases.
How can AI enhance medical affairs and medical affairs organizations, not just like MSLs? How can it improve patient outcomes? And then there was a lot of talk about efficiencies, both on the corporate side and individual. Okay. So which would you say was the most popular topic out of those? Which one was resonating the most? I think well, it's interesting because a lot it seems like everybody got excited about the efficiencies. How can I get better? How can my people get better?
But it seemed like there was also, like, a lot of eye opening information about the patient side. You know, we're in this patient you know, there's patient centricity of what can we do. It all goes back to the patient as we know. So it's like, what are some of the things I need to learn about how there's improvements in patient outcomes because of artificial intelligence? That seemed to get a lot of attention and a lot of buzz too. Okay. Cool. Cool. So tell us about some of these use cases.
Mhmm. What what were you seeing there? Give us some examples. What went down? I will. And and and this was, so it's it's this seemed to be probably the the most time was spent on this piece, And there were a lot of different speakers that covered a lot of different versions of this, but they all overlapped to an extent. Okay. So I'm gonna give you kind of a basic overview of what I saw as some of the common use cases. So number 1, analysis of information through artificial intelligence.
So data and databases and and the extracting and analysis of information so that the learnings would be streamlined and it would be more efficient or the use of data is more efficient. Also as it relates to even social media and social listening and how to extract data from social sites, and the Internet. So that's deaf that was definitely one category. Then there was the applications for data generation, which is a whole another category. Your favorite topic insights.
Yeah. You gotta tell us more there. Yeah. Well, so synthesizing insights through use of artificial intelligence was a whole another area of discussion and another use case. Then there's the communication and information sharing, tailored and personalized communication, both for internal and external stakeholders. Content development. Obviously, you know this better than anybody because you develop more amazing content than anybody out there.
You and Sarah kinda go hand and go you know, toe to toe on content creation, so I know that you guys know the importance of utilizing artificial intelligence for content creation. Training initiatives, learning and development. Obviously, as as, you know, we continue to build out training programs and the stuff that you and Sarah are doing are absolutely next level.
It's just interesting to see how important it is to use all of the benefits of artificial intelligence to not only develop the best content, but to do it in such a, time efficient manner is is really kind of the key. The other thing that was interesting too was, predictive analytics.
So looking more on the patient side of the equation, and the use of diagnostics and then, the data analysis coming from, you know, on the patient side, you know, with wearables and and and different types of new technology that has AI built into it is really, really amazing. And where things are going in science is just mind boggling. Yeah. That wow. Sounds like a ton of stuff was covered.
Do you have a granular, very specific use case you could share with the audience just to give them a a sense of, some of the details around what people were sharing? Yeah. So and the you're gonna laugh because this is I'm gonna share something that I thought was brilliant as a very detailed use case, but you wrote about it in one of your blogs or your newsletter. Trying to do a shameless plug here.
Well, the no. The this was like this literally, like, it wasn't it was so new to, I think, a lot of people there, but it it's not new to you or Sarah. It was and so, anyway, one of the things that happens with MSL specifically is just information overload. Having to educate yourself on on with on certain information and having to use that information for your KOL engagement.
So let's just say you have to plan for a KOL visit, and you wanna you wanna make sure that you're leaving no stone unturned. And maybe that KOL just released a paper, an article, maybe there's a ton of articles and papers from this KOL, maybe it's the first time you're visiting them, maybe there's just all this information that you need to curate in order for you to understand that KOL better and be prepared for that meeting.
Well, you can take all of that data, put it into notebook l m, and ask it to create a podcast that you can listen to to educate yourself on that data in preparation for that meeting. Now you know this. Yeah. People's minds exploded. Right? Exploded. Yeah. It's it's really cool, you guys. If you haven't checked it out, go to notebooklm.google. You just need a Google account, and you can upload anything you want.
So like Tom said, one paper you could even do series of papers from the same KOL, and just click a button, generate podcast. It is awesome. And if you want to learn a way to level this up, go to medicalaffairsvalue.com, find the podcast or the blog post on how to turn papers into podcasts, and I give you a specific prompt to really uplevel your KOL engagement. So definitely check that out or reach out to me if you wanna know. And now we're going back over to Tom. Alright.
So where and how can AI enhance medical affairs? What what were the folks saying in there? Yeah. So and lot of different really good contributions. I think as a summary, I wrote a couple of things down. So operational efficiencies and engagement was one whole big area. So taking, for example, manually intensive tasks, such as med info, standard response letters, patient summaries, creating animations and graphics, all through the use of AI can streamline all of that.
So, again, I think this falls into the category of, like, efficiencies, if you will, insights and personalized communication. So the the idea of tailored communications kept coming up and personalized communications, and I think that this is something that medical affairs is trying to achieve to develop and create and and build stronger relationships with KOLs. And through use of technology and AI, we can learn more about our our external stakeholders to be able to do that.
They that notebook l m example is, like, a perfect way to describe kinda how that can happen. The other thing is that was brought up a lot was evidence generation. So I didn't mention this before, but, like, one of the things that's that they talked a lot about was, you know, real world evidence and evidence generation through through AI, extracting data from databases as it relates to the patient.
Like, there's there's so much that I was unaware of and I think a lot of people are unaware of of how AI can touch the patient and really improve outcomes. We're just scratching the surface. Yeah. That sounds amazing. I had an experience myself recently on the power powerfulness. Is that a word? The power of the power of using AI to analyze insights. So, guys, never put your proprietary insights into an open model. But I just published a blog post where I took synthetic insights.
So I had AI generate a 100 insights, bad ones, and then I had AI rewrite those to make them more complete. Then I asked it to do an analysis on this. So mimicking what people in medical affairs would do, my mind exploded because it just analyzed a 100 pretty big insights in seconds. I was texting Sarah about it, actually. I was like, this would take a medical affairs person days. It's crazy. The especially the summarization capability. So, again, don't put your own insights open model.
But if you have a closed model, try it out. Ask it for a summary. Your mind will be blown. I wanna mention one other thing around personalization. So Tom is spot on here on how you can really use AI to help personalize your outreach to KOLs.
I have a blog post on this called how to use AI to uncover what your KOL values, and I have a very specific prompt that you could copy and paste into your model, whichever one you're using, and then just plug in your KOLs name, and it's gonna spit out a huge report for you. So definitely check that out as well. Love it. Let me continue to hijack this. No. Wait. This is my episode. That's your episode. You whatever you wherever you wanna go. Alright. So let's go back to the patient journey Yeah.
Stuff. You know, you were talking a lot about patient journey, which I think is really cool. I personally haven't been thinking a lot about that. But then the link to RWE or real world evidence, what tell us more about that. Yeah. So I from what I I gathered from that, it's that relates to the collection of data, and the the power of AI algorithms applied to the patient experience is a total game changer.
So how to apply treatment protocols for different diseases through evaluating and an and analyzing the data that's coming from these these algorithms that are being applied. Like, I look at it as and I don't know if this is a is a great analogy, but, rare diseases. There there are there's there are people that are walking around undiagnosed right now because they have a rare disease and they're not Nobody knows what it is.
Yeah. So just imagine how by using AI and taking that data and entering it into a system, we can I maybe narrow down and diagnose these people quicker and find treatment protocols for these people, because I don't I just don't think we're doing as good a job at that right now as we should? That's and, again, that's my that's kinda my example. That wasn't exactly something that they talked about, but when I left there, I was like, jeez. I bet you that is a thing.
That's that would be a total game changer, and I know there are a lot of medical affairs teams that show impact of their work by linking it to timed diagnosis. So they might say, oh, we have a medical education campaign on this rare disease, and then start looking at if they can, through claims and whatnot, see if the timed diagnosis is decreased and showing the impact of their work.
So I think, yeah, that is that is super cool, obviously, for showing impact, but more importantly for the patients. Right? And the other yeah. And and to answer your question too, because I did write some notes, you asked me about the, real world evidence and patient journey. One of the things that they couple things I talked about was to identify predictors from electronic health records. So that's where that, like, that whole, you know, rare diseases thing came from.
Basically, curating trial data, and how the analyzing of that data is just gonna lead to such a better accuracy in uncovering unmet needs and then inform strategy. Yeah. That's gonna be I can't wait to see more of these use cases as we go to more of these conferences. That's gonna be that's gonna be awesome. Alright. Let's talk about value. You know? Guys You're the value queen. We're all about value. Right? Adding value to you, adding value helping you add value.
So what is the value of AI for medical affairs now that you've attended this conference? Or how what do you have to say about this? Well and I took a lot of notes on this because I really felt like this was really the take like, what were the takeaways and and how how valuable is this? And it touches AI can touch so many different areas. It can define and drive strategy, HCP identification, resource allocation.
I talked before about personalized communication and the value that you can get from using AI to for messaging, for example. I talked about content development, patient centricity, real world evidence, validating insights. I one of the things I didn't mention, but this is think about the value in identifying trends Mhmm. And identifying areas of unmet medical needs.
So these are all things that I feel can come from really getting good at utilizing artificial intelligence tools and implementing them within, you know, our day to day. Yeah. That is that is super exciting. And I can see why so many people were so excited after leaving that that beating. Alright. Let's transition to AI and Metinfo. You alluded to this earlier with, standard response letters, I think. What else?
Yeah. So on the Metinfo side, what we what I picked up on is that AI is now being used, for for adverse events, specifically with detection tools. Companies are using AI for their FAQs, scientific engagement, summaries from posters, conference content, insights, and then also, you know, AI for medical writing, which is a whole another category. I kinda threw that into MedInfo, but, but definitely all these areas I kinda feel was discussed in this in the in the med info section.
Was there any specific examples there around, like FAQs or standard response response where people I guess, in my mind, and I don't know a lot about this. But in my mind, they have the AI write something, and then a human looks over it and gives it the final approval. Did they talk about anything like that during the conference? Like, any examples of how they're No. Not not specific to that. But but to your point, one of the things that came up a lot was the importance of the human element.
And the idea that this would be a machine driven process that a human runs and manages, not a machine driven process that the machine runs and manages. It just spits out and a stat and helps with the whatever that, that use case is. Yeah. That's a really good point. I always feel like, you know, a lot of people talk about, oh, AI is gonna take our jobs. And when people say that, it makes me wonder how much time they've actually spent in it.
Because when you start really playing with AI, it makes a lot of mistakes. And at this point in time, I would be surprised if it takes over, especially medical affairs jobs. What do you think, Tom? Yeah. No. I you know, it's funny. I I and look, I'm I'm hoping nobody loses their job because of AI. I hope it creates more jobs. Right now, it it absolutely requires human element. It requires well, think about it. It requires the human has to operate the system.
Otherwise, there there is nothing to generate. So if you're not if you don't know how, for example, to use the tools that you have and use the right prompts, for example, you're not gonna get the output that you need, so it's worthless.
So that's that kind of the other piece of it is I think one of the biggest challenges that I see, and you could see it by the overwhelming interest in this area, is that people still don't know what it is, how to use it, what it means for them, what tools they should use. So the reason everybody showed up, I think, to this conference and the reason people are listening to this is because they wanna see what is what what's everybody doing? What's everybody else doing? What tools are they using?
The my when you asked me about, can you give me a specific use case, and I and I told you about the notebook l m thing that I learned from you, that's a perfect example because it's a real example of how you can apply it, and that's what people are looking for. Yeah. Yeah. Oh, gosh. I had something I wanted to say, and now I totally lost it. Oh, I remember now. There's a really great article I read about comparing AI to other softwares and why the uptake is or can be challenging.
So the author equated or said that other softwares have a specific way to use them, so there's a right way to use them. With AI, there's no rules. You have to make up your own rules and really know what you want and how to get it out. And so I think this is, linking to your prompt engineering, comment. But before we go on to that, let's talk about what you learned around improving business efficiencies through AI.
That was a really big part of this, and, so what companies what what I learned is that companies are are using AI tools to improve upon metrics and goals, so that that was another piece of it. Performance reviews, reporting, definitely being able to cut down on the reporting and administrative side of the equation, call notes, I mean, you know, that seems like an obvious one, but it but is everybody really utilizing it, you know, use utilizing AI for that piece of it?
What it it I also learned is that using the right tools can make an immediate impact, and the way companies seem to be using it is by automating process workflows. So for example, a lot of people are using MS 365 power apps, which from what I understand is is a great use of both automation for efficiencies powered by AI. And it's it's kind of a low cost solution. It's it's there's no real barrier of entry to that.
So MS 365 power apps, it was mentioned a couple times, but it says that I wrote down, so if an organization has no AI solutions, it can start at the app level, with minimal investment, money, and time, and that would be that MS 365 power app would be an example. That sounds really cool because I think everybody has 365. Right? So it's an ecosystem that we're all in already. I think that sounds pretty cool.
And I I think a lot of folks, along the lines of what you were saying with business efficiencies, are not using even the features that Microsoft has. So, for example, depending on what version you have, you can just summarize emails with AI. Right. And so think about when you have those huge chains, you know, like, the really annoying ones with all the the, not signatures. I was like, out of office, no signatures. And they it just adds up and adds up, and you're, like, scrolling forever.
You can just say, summarize this email and give me action items. Like, that's, like, huge timesaver. It's such a small thing, but that adds up. That's it. But that what you just said, that's a great example. It's a great example. Yeah. Like I was there. It's you should've been there. I know. I really should. I felt that way. I know. I told them, you need to speak next year at that event. Oh, I would love to. I have a I have a at Mass West, a session coming up.
If you guys are going to that, definitely come to my session. I'm gonna give real live use cases. Oh, boy. It's gonna be gonna be really awesome. Okay. Amazing. Alright. Back to today's topic. Was there anything that was not covered that you think would have been really helpful?
Yeah. And and and this is another thing that I learned from you and from Sarah is the importance of prompt engineering, and and that really wasn't that should have been its own topic, and I feel like there there could have been at least one session on the art of prompt engineering. You guys do this better than anybody, and you have the like, seeing the scripts or the examples of prompts for MSLs that you incorporated into the training program blew my mind.
I didn't even know that that's how you do it. Like, I know how I do it, but that's so rudimentary compared to what you should do. And that that wasn't really a part of it. Yeah. Like I mentioned, AI, it doesn't have one way to use it. You don't you don't just click here, there, and there. You have to really understand how to interact with it, and having a set of rules or frameworks around that is super important.
Something Sarah and I include in our team training on AI is a whole section on writing effective prompts. So we give you frameworks on how to structure your prompts to get better output. And then going beyond that, how do you refine your output so that it's actually really useful? So I think something I've seen a lot of people do is they they kinda put in a crappy prompt, and then they get crappy results out, and then they stop. And so I think appreciating that it's iterative.
And I published an article last May on how to use AI for creativity and productivity. And I put one of my top prompts in there or refining prompts, let's call it. And I got so much outreach and comments on it. So I'll tell you at least what it is. It's if you have some AI output and you don't like any of the ideas, you know, some are okay, but you need more, you just write give me 10 more. And the AI will just spit out 10 more, and you can just keep doing this. Give me 10 more. Give me 10 more.
And I think that is a really underappreciated use case of AI just for ideas generation. It's going to eventually find something that you haven't thought of yet and just really make your creativity explode. I think that's a really cool, use case that's underappreciated. I love talking to you about this because you get so excited. Oh my god. I love it. You love this. You love this. Can I tell in the last couple of minutes my favorite use case right now? Yeah. Oh, yeah. Personally.
Yeah. So I'm a big journaler. I plan. I journal. I write a daily journal daily plan, daily journal every day. The end of the week, I take my 7 entries from the week before, and I put those into notebook lm. I tell notebook lm to act as a life coach that specializes in intentional living and to give me ideas, suggestions, prompts. Tell me how I grew over the last week, and I just interact with this chat bot to help me do my self reflection.
But to Tom's point earlier on notebooklm's capabilities for making podcasts, I click generate podcast and make a podcast about my life. And I love listening to it, and the cutest thing too my husband loves it. He's always like, can we listen to your podcast? Oh, that's awesome. It's so sweet. And it just it's so funny to hear your life described in that way. Like, it'll often start out like, oh, we're gonna take a deep dive behind the scenes in the life of an entrepreneur. Right?
And then, like, just hearing it talk about you that way is really interesting. But okay. Bringing us back to the AI Medical Affairs conference. Any other tips? So we're about to wrap up here. Any other tips or advice you can share from what you learned at the conference? Yeah. So and this this, first of all and I'm gonna coin a phrase that I heard you say last week in one of your, in one of your blog posts or your messages or post or whatever. Where are you at? Garbage out.
Yes. Garbage in, garbage out. So when it comes to prompt engineering and it comes to prompts, you really have to know what you're doing. The other thing that came through loud and clear, I touched upon this before, and this is the analogy, don't take your hand off the wheel. Oh, I like it. I like it. And the analogy is that we have self driving cars right now, but do you really feel comfortable getting in them? Like, I wanna keep my hand on the wheel. Like, okay.
Self drive, but I'm still gonna be here just in case. But no Waymos for you. No Waymo. No. No. Okay. Because Waymos are here. People use them all the time. Well, listen. I I just feel like that's a great analogy to say Yeah. You can't leave AI to its own devices, needs the human element, stay involved. I think that that's just good advice, and that was something that came through pretty loud and clear. It was talked about a lot.
And the other thing too is and I I I learned this from you, and from Sarah. One of the things I didn't realize about prompting is that you can ask AI questions and ask for feedback. How is my prompt? Should I be asking it this way? Here's what I'm looking for. Am I asking it you know? So you can actually ask for advice from the AI tool. You don't have to just sit back on your own devices and say, hey. I I need to know this. Figure this out. Yeah. So yeah.
Yeah. To along those lines, I have a ton of custom GPTs that I use for specific tasks. And people ask me, how do I write the system prompt? And I'm like, I use AI to write it. So I just tell it that it's an expert prompt engineer and that I need a a system prompt to do this. And then it writes it, and then I plug it in. It's yeah. So to Tom's point, leverage the tools, know what it's capable of. Any one last final thing, Tom, before we hang up on you? Stay on top of this.
Like, stay like, don't be afraid of it. Embrace it and try it and and jump in the pool and starts women because it's here, and the people that are embracing it, their eyes are starting to open up. They're like, wow. Jeez. This isn't that scary. I can do this, and look how much it's helping me. So I would say if you're sitting on the sidelines, get in the game and start playing around with it because it's here, and I think it really can help you. That's kinda what I learned.
Yeah. Yeah. Well, thank you so much, Tom. Let us know what you guys think of this hostile takeover of the MSL talk turned Katrina talk podcast. Let us know what you think. We always love hearing from you guys. If you're not following me and Tom and Sarah on LinkedIn, please do. We create a lot of great content, and we really love hearing from you guys. Thank you, Tom. Thanks, everybody. Your talk. I love it. It's the newest thing. Thank you, Katrina. You're the best host ever. Thanks, guys.
Bye. Thanks, guys. See you next time.