Innovation in Medical Affairs: Embracing AI for Success with Scott Thompson
33 minAudio
This episode is sponsored by Acceleration Point. Pharma Sessions is hosted by Jonathan Kaskey In this episode Jonathan and Scott Thompson, the Co-CEO of Acceleration Point, discuss the impact of Artificial Intelligence (AI) on Medical Affairs. Examples of AI use in Medical Affairs include analytics, content creation, and productivity enhancements. Scott encourages experimentation with secure, customized AI tools to improve output. He outlines a case study where AI imp... Pharma Sessions is hosted by Jonathan Kaskey Follow along on LinkedIn: https://www.linkedin.com/in/jonkaskey/ Or join the Pharma Sessions Substack: www.pharmasessions.substack.com
Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.
Transcript
From the show. Tap a time to play from there.
SPEAKER_01I tell all of my medical affairs friends the same thing I say to my 16-year-old. AI won't take your job. Someone using AI will. Right? And so we must be practicing it. We must be learning this new technology and learn how we can get the most out of it.
SPEAKER_00Hello, hello, and welcome to Pharma Sessions, the podcast where Metaffairs professionals can learn from each other. Today's episode is sponsored by Acceleration Point. What does artificial intelligence mean to Metafairs? How is Metaffares using AI today, and what is the promise for the future? And how can understanding this help you thrive professionally? In this episode of Pharma Sessions, we talk specifically about AI and its impact on Metafairs. My guest, Scott Thompson, co-CEO of Acceleration Point, has deep expertise on how AI is transforming the industry.
SPEAKER_00From analytics to content creation and productivity, Scott discusses the various ways AI is revolutionizing Metaffairs. So let's dive in. So today we're going to talk about AI and medical affairs, which is a huge topic. So before we even get into all that, can you just tell people a little bit about who you are and what you're bringing to this conversation? Sure.
SPEAKER_01So I'm Scott with Acceleration Point. We've been working with medical affairs teams for about 10 years, helping them learn from uh physicians around the world, uh really focused on insight generation and really enjoy our work and love to talk about AI today because it's certainly part of that.
SPEAKER_00Excellent. And this I'm just curious because 10 years ago is about when I started looking into medical affairs as well. And I didn't know a heck of a lot about it before that point. So did you know of MetAffairs beforehand? How did you get in this realm in the first place?
SPEAKER_01So 10 years ago, I had actually was in the space of organizational effectiveness and worked across multiple industries. And 10 years ago, started working with Biogen, specifically in the area of technology and uh process improvement in medical affairs. And from there, we went from Biogen to Sanafi to GSK, Navardis, Gilead. We've had quite the run and really been our our key has been how do you not implement technology for technology's sake, but to actually help medical affairs achieve their goals and communicate their impact.
SPEAKER_00I think it's really interesting because most of the people that I have interviewed on this podcast are actually working for the biotechs or the pharma companies themselves. But I always think it's interesting when you talk to somebody like you who's had that like vendor's perspective of you just listed off five or ten different companies, right, that all have their different ways of doing things. And I'm sure some are good and and some are bad. But I also think it's just a very helpful perspective for people because that's a career's worth of experience, right? That you rolled off in a minute.
SPEAKER_01And we also have folks from industry that that come and work on our teams. Our VP of medical excellence has been in the industry for I'm not allowed to say how many years, but yeah, she's been in for her entire career, and it's been great to get to work with her too.
SPEAKER_00So when did people start coming to you and talking about AI? Obviously, it's like the huge topic and uh all over everything right now, but was it prior to this wave of interest? How did that sort of start?
SPEAKER_01So AI isn't actually new. AI has been around for 15, 20 years, uh, but it hasn't been as accessible uh to the everyday person since this most recent iteration of large language models made famous from Chat GPT, and now you see it embedded in everything. So AI has been around. I guess the first ask around AI was probably eight years ago uh when we were working with uh a large medical affairs team that wanted to uh do work in what they called big data.
SPEAKER_01Um they wanted to take uh disparate data sets, um, different kinds of data sets, everything from claims to uh publications and conference data and trial data and try to put it all into a machine and let that machine learn it and then provide insights that you couldn't see with the naked eye because it was just too much data to look at. So that was back eight years ago.
SPEAKER_00That's yeah. What's funny? I've heard some people say that AI is how people refer to technologies in the future state. And then when it starts to get applied, you're like, oh, that's not AI anymore, that's Google Maps, right? That's my navigation. That's something I use all the time. AI is in the future. But we're already using AI, right? And this I think you immediately jump to it, but the idea of AI for processing big data, AI replacing big data. Do you see these as two different flavors of the same approach, or does it even matter how you draw that distinction?
SPEAKER_01So I I think that it it's not useful to try to separate them because big data would be the big data sets. AI is one of many methods to interact with that data. Right. But I think for medical affairs in particular, the big shift is that what would have normally taken a team of data scientists months to prepare for you can be put in your hands and you can do yourself without all of that help. And are there differences between what data scientists can do and you?
SPEAKER_01Of course. But you know, the vast majority of what we want to accomplish using it, we can actually go in and do without that big team.
SPEAKER_00Yeah. And when you say go in and do, because one of the one of the conceits of this podcast is that we're trying to get to actionable information, right? That somebody in Metafairs can hear and then maybe do the job a little bit differently. At least that's a goal. So what are you seeing for how Metafairs organizations are currently using AI?
SPEAKER_01So I guess I'd put AI into three buckets. So the first bucket is analytics. And this is usually where most people will come to us at first saying, hey, I we're interested in AI, and we say, what are you interested in doing with it? And they come in thinking about analytics. And this is uh putting AI on your uh insights that are collected from the field. It might be uh using AI to process the vast amounts of uh social media data when you're looking to find out what HCPs are saying in the public domain.
SPEAKER_01We do both of those in our product, but there's also teams that are using different AI products for that. It might be to identify study sites, right? So if you know what the patterns and trends are that you're looking for and you have that data, you could ask it questions to find things that match that pattern. So that's what I think of in analytics. Uh the second big use is around content. And so there's a lot of experimentation right now on how you can build plain language summaries, how you can create interacted to help with MedInfo and other omni-channel inclusive pieces.
SPEAKER_01And so, how do you create that content in a way that it can be interacted with through AI? And then the last one, which I think is underrepresented, but is where I would start, is productivity. I can't tell you how many hours a week I save because I use AI to help with everything from emails to presentations to analysis in Excel to writing thank you notes to my kids' teachers after an event. I just use it. It's become just like as useful as PowerPoint or Word when you start to use it that way.
SPEAKER_00There's wow, that's very interesting, right? Because and I like how you've broken it up into the three buckets, because that kind of makes sense even as you were describing the progression of people coming to you and talking about projects. Like the first project that you described was in that analytical realm of we have all these different data sources, and then I feel like around insights, you have you had advisory boards, you have MSL teams, you have people posting on various monitoring academic journals, you have social, right? So that is so natural that you could say, hey, I want to, it's really important to me that as a medical affairs leader, I want all of my MSLs to to know as much as possible about the doctors they're about to go see and to go in, even if that means that person had an interesting comment on an advisory board three months ago, or they just liked something on on Twitter on X or whatever.
SPEAKER_00So that all seems that all makes sense. And then the content, one thing that I've seen working in the content space is there are sometimes these directives that come down from the pharma organizations and this one particular big pharma that I was working with, they basically had a directive that's don't put our proprietary stuff on Chat GPT, we don't want it learning on it, et cetera, et cetera. But you specifically mentioned plain language summaries. And what they found was that because they weren't giving these plain language summaries fast enough to the people that are interacting with with patients, the sites themselves were actually going and they were saying, Hey, give me a pay I'm putting this in chat GPT, summarize this for me.
SPEAKER_00So I just thought it was fascinating. They tried to prevent it with a corporate directive, but people outside their control just did it anyway because we're just getting there so fast, right?
SPEAKER_01Yeah, it's really one of the fastest upticks in technology we've seen, right? The amount of time it took for people to get online and then get high speed and then get on Google, and now to move to this, it's just moving so quickly. Either you're going to learn about it and adapt to it, or you're going to get left behind. It it's something that is just part of every tech that we're in now.
SPEAKER_00So that's a fear, right, that a lot of people have. There's the straight out, is AI gonna take my job fear? But then I think another thing that when people really think about it is what's the danger that somebody who's really good at AI takes my job? That might be even more of a risk, right? So, how do people get involved in this and get started?
SPEAKER_01First of all, I tell all of my medical affairs clients the same thing I tell my 16-year-old son, AI is not gonna take your job. Somebody using it will. And it's about become and it's not just AI, it's staying up to speed on recent technologies, recent trends, finding ways to be more productive, asking better questions of the data. Because here's the thing in medical affairs, nobody that I know got into medical affairs because they love technology and databases and language model training.
SPEAKER_01That's not why we're here. And AI won't take that person's job because the key to AI is knowing what questions to ask so that you can get something useful tied to your strategy. And so as long as you're focused on building a high-quality insight strategy, you know what your scientific journey is, you know what your listening priorities for, you're still going to need to do all of that good foundational work. Because whether you ask a team of a hundred analysts to go find you an answer, or you ask a computer to find you the answer, if you don't know what you're asking for, you're gonna still get bad stuff back.
SPEAKER_01Do that better because it can happen faster.
SPEAKER_00I love that. I love that, right? And it's almost it almost makes me think back to the the last kind of big new channel that was so hot was social, right? When social first came out, whenever, whenever that was. And eventually I think people got very focused on the channel itself, and then they shifted back and they're like, oh no, we actually just need to focus on our particular goal, and then we're choosing we're choosing the right medicine to solve the problem. What that might be social, that might be traditional, right? And AI, what you're talking about is you have a question to ask and you're trying to find the fastest way to get the answer.
SPEAKER_01Yeah, like in Quello, which is our product for for insight generation, whether the data point is information collected from the field medical team, or if it's we're listening to thousands and thousands of HCPs and what they're saying online, if that's your data source, if you're looking at ad boards, you want to you need your data sources and you need to be harvesting them and maintaining them. And then AI sits on top. And AI can do a ton of things, everything from summarizing to tagging, reporting.
SPEAKER_01Um, I think one of the most uh interesting things that uh a lot of uh folks don't remember is that they'll always use AI on the back end to look at all that data, but it also can be used on the front end to make sure that the data coming in is cleaner, stronger, more precise before you even analyze it. So we're gonna start to see AI get embedded in more parts of the process, not just the analytics.
SPEAKER_00Interesting. So currently, you're saying the medical affairs approach to AI is focused around analytics, content, and then productivity. What do you think that these companies are not doing yet that they should be doing?
SPEAKER_01So I don't know if they're not doing yet. So first, I I think the big thing that they're not doing that they need to do is experiment. One of the things that medical affairs teams, pharma in general, but medical affairs teams even more is known for, is uh not getting started until the full plan is in place. Right? Unless you have every T crossed, every I dotted, you've tested it a million times, it's really hard for medical affairs to roll out new technologies. Where this set of AI technologies is iterative.
SPEAKER_01Right? You can roll something out and then roll the next piece out, and then the faster and better you get at it, the better it gets at you. So I would start encouraging your team to experiment, but within secure protocols. You mentioned earlier.
SPEAKER_00Yeah, talk more talk more about that.
SPEAKER_01Yeah, you mentioned earlier this fear because we've seen it on in the news, where if you start to use Chat GPT, it's gonna start learning from your data. Does that create a security risk, a competitive threat, things like that? So there are ways of using what I'll just call this GPT technology in models that are secure. They're called ring fenced. And what that means is it two things. One, the model doesn't learn and share that learning with other people using the model because it's yours.
SPEAKER_01And two, you can direct the knowledge that it is using to generate answers to a set of data that you specify. For example, let's say I had a hundred uh insights in my system, but then there's a million other data points on the internet. You don't want it to balance those, you want to just say across these 100, what are people saying, right? Because I don't know who generated the other million. And if you just leave it out there, it's not going to realize how to balance those.
SPEAKER_01So you want to have those set up. There are very specific models. But here's the thing don't wait until it's a fully baked corporate program with a hundred steps in it to get it out. Get it stood up and let people start working on it, right? They don't need to be trained even, right? You just you start getting access to this and you learn by using it.
SPEAKER_00That's fascinating, right? Is you're saying essentially empower your teams with almost a not a sandbox environment because it's actually working, but it's real data. But perhaps there are activities that are other than learning and getting answers, you're not creating things that are being published outside your organization just yet. You're ex using this as a new way to exam data, and in theory, you're learning about how do you do things like write prompts and and the like. Yeah. Do you have what just out of curiosity for as far as prompting questions, right?
SPEAKER_00Because you you had alluded to earlier, you yourself using this for general productivity things like emails and presentations and analysis. And I'm hearing that, and I'm like, whoa, that might be something that isn't dealing with proprietary data where people might have a little bit more leeway to to experiment. So if you're just if you're listening to this and you're like, hey, I'm an MSL or I'm in the field or I'm in meta, I'm in Metaffairs, but I'm not specifically tasked with developing our AI strategy or whatever, I don't want to wait for a ring-fest environment.
SPEAKER_00Where can how would people get started if they're trying to?
SPEAKER_01So I would say you should have Chat GPT on your phone and on your non-work computer. Right? And you should start using it. You should start using it for draft me an email to my team asking everybody what they would like to do while they are in uh Las Vegas at our upcoming meeting. You'll get a really nice email. It would take two seconds, and then it'll you can have it brainstorm ideas. You could say, Hey, I'm I use it a lot for could you craft me an email?
SPEAKER_01So I'll say, I want an email to Jonathan. I really like to be on his podcast. There's three things that I like about it, and could you send him this? And it will create this really nice worded email, it'll cite the reasons I'm interested, and then I just copy and paste it in. And it just saves you a lot of time. And more importantly, you start to learn how to use it so that when you do want to do something more sophisticated, you can um be familiar with the with the protocol, with the the rhythm of asking iterative questions.
SPEAKER_00This episode is sponsored by Structured Meetings. It's a Teams app that turns your own Teams incidents into the perfect advisory board platform to save money, get insights faster, and keep your data secure. Check it out at www.structuredmeetings.com. So you're saying so you're using it you're separating it where for some of the just the general I'm planning my team meeting or whatever, right? That is a perfect use case for chat GPT.
SPEAKER_00Find me, find, plan our itinerary, right? Where are we gonna go to eat and what should people have? And let's come up with that. And then for presentations and Excel, because that's the other thing that has a lot of appeal, just also me personally, right? Like I do presentations all the time, but I'm not super confident that I'm the world's greatest PowerPoint designer. Or have you had any success using AI or GPT programs to create materials?
SPEAKER_01Yeah, so there's different tools that are better at different things, use cases of GPT. But if you use Chat GPT, you say, I need to create a presentation on the best practices for insight generation for medical affairs teams. Could you outline a presentation for me? I've got 45 minutes, and it will create ideas. Now, they might not be great. This is the whole idea of GPTs, and you go, hey, this is great, but can you spend more time on how to ask good questions to collect insights data and less time on the corporate process of prioritizing what to listen for, right?
SPEAKER_01And it will change the script. Now you've got your outline. From there, there are tools that you can say, great, make me a slide for each of these, and it will, and some of those are free plugins. Uh, if you've got Microsoft Copilot, it's built right into PowerPoint now. Yeah, and it'll do that. Uh, you'd also mention Excel, right? A lot of people just aren't up to speed on all the formulas and right. And so now you could actually say, I have this data. Um, I need in cell B56 a weighted average across months of this, and you can just tell it what you want, and it'll say, copy this formula, put it in.
SPEAKER_00Right. That's it's amazing. Uh so the copilot, I was thinking that would probably be the shortest lift for anybody to begin playing with it, is if you can get that turned on your corporate decision though versus the individual decision. Sure, sure. So tell me, Scott, about any medical affairs teams that you're aware of that are doing this AI really well now and how they got there.
SPEAKER_01We were working with a global medical affairs team across, I think it's 14 indications. And uh I think that they are certainly ahead of the curve. But I think that it's important to think a bit of where were they? What was already in place that made that happen? What are they doing today? And maybe more importantly, where are they heading with it? What's coming soon? First, it doesn't make sense to use technology to make something go faster if you don't know what you're trying to make go faster.
SPEAKER_01Right. And so this team was already, they had very strongly uh created medical plans, a very clear scientific narrative. We worked with them to help take that scientific narrative and turn it into what we call a scientific journey, which is a path that you're trying to bring the medical affairs community down to align with that narrative. And if you don't know that sequence, it's really hard to train a computer uh to go uh recognize that. So that's part one. You have to have the foundations in place.
SPEAKER_01If you don't crystal clear, if you don't have your scientific journey documented and you don't have how you want teams to work together to achieve it, it's really hard to train computers to help you do that. Okay. So have the foundations. Where they are today is really three things. The first is put AI in the hands of your analysts. So it's still a human using AI to help them identify data and patterns and trends faster and more complete than you could do on your own.
SPEAKER_01The second is that they were putting they're putting AI into the hands of every MSL. And the way that they're doing it is by using a digital assistant within Quello that basically, as I'm writing an insight, as I am submitting something into my system, it's checking it for quality, for references, for compliance, right? It's checking because then it's increasing the quality of the data that the analysts look at later.
SPEAKER_01And then the third is that they're doing this across multiple channels. So they're using one AI model, one platform, and then looking at what is my field submitting, what am I seeing across claims, what am I seeing in my scientific profiles for publications and Congress activity, things like that. But it's based on that one process. So that's what they're doing today with it.
SPEAKER_00That's fascinating. And then what are they what's what are they gaining from this? What's the outcome?
SPEAKER_01Yeah, so there's a handful of things. So one is they're dramatically increasing the speed in which they can look at data sets. Dramatically increasing it. They're turning, essentially, when they have a question, they're able to get a targeted answer versus having to sift through all that same data again to get that answer. And then the third thing is that the quality of the data has gone up, right? When you read the insights that were submitted before, and then you read the insights that were submitted afterwards, they all have a what somebody believes and why do they believe it.
SPEAKER_01They all have a, they're not written six paragraphs when five sentences would have told the same thing, right? You started seeing less things that are irrelevant. Doctor's going to be at this conference, right? And it's because it's built into that moment of submission versus having it correct it later.
SPEAKER_00And that is what I think you had mentioned that earlier, right? Of seeing AI start to actually impact the quality of the system going in, or quality of the information going in that then is used later as part of a data set that same I AI, different AI, is now parsing. So it's almost like you're the MSL, your digital assistant is training them to think or to write like a programmer of these things that can then be parsed out later.
SPEAKER_00How did was it a big change for the team? How did they react to that when they started you?
SPEAKER_01At first, there's always fear that that it's not going to, or that you're at first always set up as hey, I don't want to be told what to say. I don't want to be told what to write. This is a digital assistant, not a MSL replacement. So it helps you write more concise. It helps you say what you mean. It doesn't know what you mean. What we're finding is that after people start using it, they really start to appreciate it and they wish it was built into Outlook and they wish that it was built into work.
SPEAKER_01But it's really important to set it up as a tool that helps you, not something that does it for you.
SPEAKER_00Yeah. And that's I've seen a little bit of that and some of the work that that I've done with companies on, I think you'd originally, or a while ago, you alluded to chatbots and med-info, right? And I know FactMI has this program of having this centralized MedInfo chatbot experience, which I've been involved with a little bit. And it's really interesting. What you just described, we see all the time, where it's anytime you're asking somebody to change the way that they might have been doing things for quite some time, there's a natural resistance.
SPEAKER_00But like any change management, when people start to understand the why of it and the impact that it could have, there's an element of it, or it's hey, if if you can do this, you get to spend a lot less time doing the boring parts of your job and a lot more time doing the next level thinking of your job, which is why we're all here in the first place. So it can't be helpful, right? If you can position it.
SPEAKER_01And if you start smaller, like we've talked about a few times, and just get people used to it, it builds trust. The user starts to trust the system that it understands me, that it's providing me real answers that I can count on, and the system starts to learn. What this does is it will set up the next phase. It'll set up where this is all heading. So imagine a world where that same uh engine is now recommending actions back to the team.
SPEAKER_01Hey, here's who you should engage with and what you should talk about. Hey, here's three new HCPs that just came into the radar that are going to be um impactful in three years. You should start uh interacting with them. And so this whole concept of next best action, omni-channel coordination can all be started to uh you can start to utilize the same models to initiate that. And I see a world soon where every MSL complains about their CRM.
SPEAKER_01Imagine getting into your car, pushing a button, and it says, Jonathan, who did you meet with today? I met with Jonathan from XYZ. Oh, great, what did you talk about? And you just have a conversation with your car and goes, Great, I'll put all that in the CRM for you.
SPEAKER_00Exactly. Your life, you just got one, one, one free login of Viva back in your day. No, it's very cool. And I think because then when it is surfacing that and it's saying, hey, these people are up and coming based on X, Y, and Z, you as MSL still get to use your creative thinking and your brain, and you understand, okay, I just might be having better conversations because I'm going in, and not only do I know uh all the standard stuff, but I also know but this algorithm is or this AI tool is telling me that this person's patient population might be much more concerned about access than anything else, right?
SPEAKER_00Because that's who they serve. And then your conversations will go better because you just have this prep going into it. Yeah. Very interesting. So do you think so 10, let's say it's 10 years from now, right? We're in the year 2034. Do you think that the day-to-day of a say a field MSL person is going to look significantly different? Do you think it'll be mainly the same? How much do you think it'll be impacted by all the AI stuff we've talked about today?
SPEAKER_01So I think that 10 years from now, we are going to see that medical affairs has a dramatically larger impact and influence in the medical community than we do today for two reasons. One, our reach will be significantly increased by leveraging these technologies, right? Once the trust is there, once we are sure they understand us and it understands you, you'll be able to do more with your day, right? Because you won't have to hold those 30 people that are on my KOL list in my head and manage what are they saying, what are they doing, I have to focus so much on those.
SPEAKER_01We're gonna be able to continue expanding to other segments, other parts of the medical community that are important to uh patient care. I think the other thing that's gonna be really big is that we're going to see that AI helps us communicate the impact that we're having. You know, medical affairs is having a difficult time communicating and measuring the their impact. Uh not you know down to a number. It's not I increase. But what we will see is that if we can continue in medical affairs to focus on communicating and educating on data and on science, clearly this is going to help us do that better.
SPEAKER_01We're gonna think of new ways, we're gonna make it more concise, we're going to say things better as a result. And then from an insights perspective, we're going to be able to bring in more insights, we're going to be able to uh process them faster, and we're going to be able to more quickly derive new actions we can take.
SPEAKER_00I love it. All right, let me be the first to formally invite you to come back 10 years from now. That was great for the first episode of season 12. Excellent, Scott. It's been a pleasure today. Thank you. Take care, Jonathan. This episode is sponsored by Structured Meetings. For listeners who don't know my background, I'm a huge fan of asynchronous adboards. I used to work for the leading provider of async adboards, and, not to brag, but my clients had in more events than anyone else's.
SPEAKER_00I've literally led thousands from the vendor side, so I can humbly claim some expertise. One of the questions I used to get all the time was, can we just do this in teams? And the answer back then was, not really. Teams out of the box is clunky and not fit for purpose. Structured meetings change that by creating an MS Teams app that lets you have these discussions in a really smart way in your own teams instance. It changes teams just enough to check all the boxes for what's needed for great ad boards. And also it's super fast and it saves people a ton of money.
SPEAKER_00Visit www.structured meetings.com for more info. All right, well, back to the show.
Dr. Paul Peter Tak, President and CEO of Candel Therapeutics, sits down with Jonathan to discuss the company's sprint from positive Phase III results to commercial launch. They explore how Candel is redefining the...
In this episode, Jonathan sits down with Annel Duckels, who leads worldwide commercialization learning at Bristol Myers Squibb, to talk about modernizing pharma training through data and AI. Annel's unconventional path...
Pharma Sessions - ACTO Client Summit: Hallway Interviews At ACTO's 2026 client summit in Boston, we pulled some incredible minds out of the hallway for quick, candid conversations about where field excellence, AI, and...
What happens when you put blockchain underneath AI in healthcare? According to Archita Samant, Head of Data Strategy & Analytics for Global Vaccines at Merck - you finally solve the interoperability problem that's been...
What separates a good pharma sales rep from a great one? According to Jennifer Muszik, it's the ability to understand the question behind the question - the real need behind what a customer is asking. In this episode,...