00;00;00;00

MAPS

Oh. Welcome to this episode of the Medical Affairs Professional Society podcast “Elevate”. The views expressed in this recording are those of the individuals, and do not necessarily reflect on the opinions of MAPS or the companies with which they are affiliated. This presentation is for informational purposes only and is not intended as legal or regulatory advice. And now for today’s “Elevate” episode.

00;00;33;09

Jennifer Riggins

Welcome to Elevate the Medical Affairs Professional Societies podcast as a series. Within this podcast, we focus on digital first communications, how digital is transforming medical affairs. This is where we explore how emerging technologies are reshaping the landscape of scientific exchange. I’m Jennifer Riggins, your co-host. For today’s episode.

00;00;56;20

Steve Casey

And I’m Steve Casey. Today, we’re driving into a topic at the cutting edge of medical information. The use of generative AI gen AI for content creation in medical information. Specifically, we’re talking about real world pilots moving to scale. What does it actually look like to implement, again, AI solution in a medical information department. What are the risks? What are the rewards? And what should others learn from those who’ve already tested the waters? With us today are two industry leaders who’ve been directly involved in these efforts. Stephanie Vezina, Director, Medical Information at Pfizer, and Katarzyna Dabrowska, Medical Information Content Lead at Pfizer.

00;01;39;24

Jennifer Riggins

Thank you both for joining us today. Stephanie, let’s start with you. Can you give us a high level overview of your gen AI pilot?

00;01;48;02

Stephanie Vezina

Sure. So we kicked things off in early 2024 with a small experiment. Really? Just testing out summarization using an internal JNI tool. It was exploratory, but the results were promising enough that we actually decided to go bigger. We pulled together a team of four colleagues and explored three use cases literature searching, summarizing different types of content, and extracting information from product labels to help us create patient responses. That particular experiment really opened our eyes to what a Jenny I told build specifically for medical information could do. Then you fast forward to January 2025 and we started building our own internal Jenny I tool tailored format. And so it was a true collaboration between our I experts and the digital and development teams. By June this year, we had a minimum viable product, what we call our MVP. And that product was in the end of a small group of content creators. It’s been really exciting to see it come to life.

00;02;57;00

Steve Casey

Your team has done some really nice work there, Stephanie. Thank you for sharing it with us. Cassie, let’s get a little more tactical. I believe we all know by now that AI needs data to properly work. Can you give us an overview of what you guys did, if anything, to prepare your data for the AI implementation? Did you integrate Gen AI into your existing workflows, or did you have to change your workflows?

00;03;20;20

Katarzyna Dabrowska

In fact the first step was about helping the team understand what kind of data could be used with AI, especially from a copyright and size perspective. We did a lot of upfront communication and training around it. Once we committed ourselves to building our own tool, we worked closely with the development team to define the types of data we’d need, like clinical study reports, posters, abstracts, publications, and product labels. We ran several proofs of concepts to test how different models handled the data, and finally chose the one that gave us the best quality output. And just to be clear, there’s always a human in the loop from selecting source documents to navigating through the two, prompting, reviewing, and approving content. Our colleagues are involved every step of the way. The tool supports the process, but it doesn’t replace the colleague using it.

00;04;13;13

Steve Casey

I think that’s very helpful for the audience to understand. Just to follow up on what you have shared. How did you advance the use case? Beyond the proof of concept stage? And what are the steps are being taken toward broader implementation?

00;04;27;17

Katarzyna Dabrowska

Well, we are now at the point where the tool helps us with first draft content generation. But getting here to four key phases. First we had a discovery phase. Our digital and development teams met with medical information experts to really understand our workflows. Then came the design phase where we focused on making the platform user friendly. After that, we went into development sprints with medical information experts involved throughout week of testing. And they’re finding the tool, especially the large language model known as LLM, behind it. And finally, we had an extensive user testing period before launching the MVP. We brought in colleagues from around the world who are involved in content creation to make sure we got it right. Currently, we are at the stage where the tool can support us in parts of our first draft creation process.

00;05;21;00

Jennifer Riggins

All right. So that’s really cool. I think it’s great that you’ve made this much progress. I assume you initiated this work with a small implementation team. And now that you’re scaling, how have you gone about getting the rest of the content team on board? And what did the training and the rollout process look like?

00;05;40;14

Katarzyna Dabrowska

Well, we started small with just a select working group on US content. Most of them had already been involved in testing or had used internal AI tools before. So the learning curve wasn’t too steep. I led the initial training sessions in person, followed by two more technical online sessions. We also developed a set out and set up SharePoint site with training materials, and initiated office hours and communication channels to support users. Now we are identifying subject matter experts from other therapeutic areas to become trainers and champions as we prepare for a potential global rollout.

00;06;20;17

Steve Casey

What about things like governance, compliance, collaboration with other teams like it, and getting, and keeping buy in from medical leadership. How did you tackle Those items?

00;06;31;24

Stephanie Vezina

Well, actually, at Pfizer, we follow a structured project approach that brings together cross-functional teams with a clear leadership and collaboration. I led this particular project, working closely with our leadership sponsors to provide regular updates, secured a necessary funding, and proactively addressed any challenges that arose in time. Our crew included colleagues from digital Am I process efficiency and operation, which really made collaboration much, much smoother. We also worked closely with legal and our copyrights team to make sure everything was compliant. And beyond leading the tool development, I also focused on changed management, helping colleagues get comfortable with Genii in general and developing the skills they’ll need once the tool will roll out to all. I do think that communicating back to the leaders, the overall progress our global colleagues were making and using other Genii internal tools in their day to day, served as a great motivation to keep the ball rolling with the development of this specific tool, because they saw that colleagues were actually preparing to be bold and skilled at using Genii in the future.

00;07;47;05

Jennifer Riggins

Yeah. I think it’s really great that you were able to include so many functional areas. In this process. And change management is always so important. As we all know way too well, you can never overcommunicate either. So how did you make this work? How has this fit into your overall strategy and what benefits have you seen and what are you still hoping to accomplish?

00;08;12;10

Stephanie Vezina

Yeah. So our big picture goal is to make medical information more accessible and useful, right when and where healthcare professionals need it. The Jenny I tool helps us get there by speeding up the first draft process and the quality of those drafts is instrumental to those efficiency gains. That time savings and the increased operational efficiency means that our team can focus more on strategic work, such as collaborating with medical colleagues, deepening their product knowledge, and really becoming thought partners. We’re still rolling the tool out globally, but the feedback so far has been encouraging and we’re continuously refining it. Like Katarzyna mentioned, based on what our early users are telling us.

00;09;03;06

Steve Casey

You know, I’m just wondering, were there any unexpected positive outcomes?

00;09;07;09

Katarzyna Dabrowska

Oh, absolutely. Steve. One thing that really surprised us in the best way was how quickly people embraced AI in their daily work. I think a lot of that comes down to the adoption plan that we put in place. We didn’t just hand people a tool and say, good luck. We gave them clear goals, a structured training curriculum and ongoing support. And it worked. I mean, we saw a remarkable uptick in how often people were using other internal AI tools. Now we even have colleagues proactively asking for access to our summarization tool. That kind of enthusiasm is a great thing.

00;09;43;29

Jennifer Riggins

I love hearing that. I think it’s going to be so important that people just get really comfortable using these gen AI tools. But let’s talk about the bumps in the road. We know there are always bumps in the road. So, Stephanie, what were your biggest challenges?

00;10;00;18

Stephanie Vezina

Oh, there were definitely some bumps along the way. One of the biggest bump was just helping other teams, especially our tech team, understand what medical information actually does. Decision making around which features would be included in the tool involved extensive cross-functional discussions. These conversations were actually essential to align on the medical information needs, understanding the existing processes, and to assess the capabilities and limitations. The tech team needed to be aware of while that process was complex, it ensured that decisions were well informed and balanced across all the stakeholder perspectives. We also ran multiple proof of concepts to make sure we were using the best possible Lem. That added time and complexity. But it was actually worth it for the quality we’re seeing now. And of course, securing budget for something new before you can fully show its value is always a bit of an art. We make the case, and, fortunately enough, we’re seeing the pay off now.

00;11;14;14

Steve Casey

You know, we hear a lot from other medical affairs professionals that management, you know, senior management wants them to use AI. What advice would you give to fellow medical affairs professionals looking to use AI in their operations? Let’s start with you, Katarzyna.

00;11;30;28

Katarzyna Dabrowska

For my experience, if you’re thinking about bringing AI into your operations. My advice is to start by mapping out your processes. Understand the bigger picture. Then break it down into feasible chunks. And don’t stop training on the tool. Support your users with office hours, workshops, newsletter, whatever it takes. Having subject matter experts available to guide users makes a huge difference.

00;11;57;17

Stephanie Vezina

And I think I’d also say, Startup. Stealing your team early ensured they understand Jenni’s capabilities and limitations, as well as any relevant internal, corporate or external policies they would have to work with even before the tool is ready. Maybe Anchorage small daily uses of AI to build comfort and confidence and also address the fear factor head on. Jenny, I isn’t here to take jobs. It’s here to help us do our jobs better and faster. Then the real question becomes, what will you do with the time you save?

00;12;36;29

Jennifer Riggins

I like that. So what’s next for your programs?

00;12;40;15

Stephanie Vezina

While we’re keeping a close eye on how the technology evolves and it evolves fast. That means, refining our current tool or even maybe switching to a more advanced platform down the line. At the same time, we’re continuing to focus on change management, making sure our teams are not just equipped but excited to use these tools.

00;13;02;12

Jennifer Riggins

Awesome. Any closing thoughts from you, Katarzyna?

00;13;05;28

Katarzyna Dabrowska

Certainly. Jennifer. Jen, AI is changing the game for medical information. It’s not replacing us. It’s evolving our roles. We’re adding new skills like prompting and digital fluency to our toolkit. And that’s exciting. It’s a chance to grow and lead in a space that’s moving fast, and we need to evolve with it.

00;13;26;29

Steve Casey

You know, from this conversation, it’s obvious to me that Gen AI is starting to move from kind of, quote, that interesting technology on quote to real operational value and medical information. But I think as we move from POC to implementation, there’s a need for a more holistic strategy to ensure the best results.

00;13;47;20

Jennifer Riggins

Yeah. I couldn’t agree with you more, Steve. I think really, getting that strategy around everything and understanding how to. How do you how do you flush it out across the board? Get that change management in place, make our employees feel comfortable and confident in the new technology and get that holistic strategy. And that does lead to the best results.

00;14;12;02

Steve Casey

Absolutely.

00;14;12;29

Jennifer Riggins

Thank you both for sharing such detailed and practical insights. I think this kind of open dialog is how we definitely will be able to move the field forward.

00;14;22;04

Steve Casey

And thank you to our listeners. Be sure to follow Elevate Digital First Medical Affairs wherever you get your podcasts. Until next time, stay curious and stay innovative.