Unlocking AI’s Potential in Pharma: A Conversation with Katarzyna Stoltmann & Lea Grotenrath
Wed 28 May 2025
As the healthcare sector moves deeper into the era of AI transformation, pharma leaders are navigating not only data complexity but regulatory rigour, real-world impact, and the cultural shifts required to embed AI at scale.
Ahead of their joint keynote at Big Data & AI World Frankfurt on 4-5 June, Katarzyna Stoltmann, Head of Data Science & AI Solutions at AstraZeneca, and Lea Grotenrath, Manager of Real-World Data Science at AstraZeneca, share their perspectives on how AI is reshaping pharmaceutical strategy—from infrastructure and talent to insights and patient outcomes.
In this interview, Katarzyna draws from nearly two decades in data science and AI leadership, while Lea brings a real-world, public health-informed lens to how data can improve care. Together, they discuss the human, strategic, and scientific factors that will define the next phase of AI in life sciences.
– – – – – –
Katarzyna & Lea, you’ve chosen to speak at Big Data & AI World Frankfurt. What inspired your decision to take part, and what themes are you most passionate about exploring together?
Katarzyna: First of all, Rebecca, thank you very much for having us here. It’s a pleasure to be here. It was a great pleasure to be a keynote speaker at the previous Big Data and AI World in Frankfurt. As a leader in data science and AI, this event is essential for me since it’s one of the largest tech forums in Germany. So when Ivett asked if I would like to be a keynote speaker again this year, it wasn’t a question of if but rather which topic I would like to cover. Should it be more technical or more strategic? Thanks to Lea, we will address both.
Lea: Thank you very much for also inviting me. I’m coming from a public health perspective, so attending such a big conference that brings together so many different industries is a unique opportunity. I hope to learn from other experts and get an understanding of different perspectives. I think I will go home very inspired.
Lea, your work sits at the intersection of data science, medicine, and public health. What excites you most about this space right now?
Lea: So for me , it’s really connecting both worlds. So on the one hand we’ve got healthcare system and healthcare provision and it’s vital to have an understanding what are the greatest needs. On the other hand, it’s very great to have an understanding of data, data science and its implications, which is required for deploying qualitative solutions and also ones that meet the requirements of different kind of stakeholders. So my role is really trying to connecting both worlds by building bridges.
Lea, what role does real-world data play in accelerating meaningful AI adoption in healthcare and life sciences?
Lea: Real-world data in healthcare comes from a wide variety of sources — for example, electronic medical records, billing processes, registries, and even directly from patients through patient-reported outcomes. Every minute, every day, new healthcare data is being collected. The volume and variety of real-world data in this field are immense.
Here in healthcare, and specifically in Germany, we face several challenges, such as limited data access and issues with data interoperability. Moreover, real-world data is often secondary data, meaning we lack access to real-time, primary data. This is a crucial issue that needs to be addressed when applying data science and developing AI solutions. In general, when working with real-world data, just like with any other data sources, we need high-quality, unbiased data. That forms the foundation for effective data science.
Katarzyna, what do you see as the biggest opportunity for AI to positively impact enterprise in the next 2–3 years?
Katarzyna: I love the focus on positive impact — I’m an optimist, so I really appreciate that. It’s a tricky question, given the many external factors beyond our control and the impactful events we’ve experienced globally. But it’s important to keep a positive outlook, as you’ve framed it. Ideally, the term ‘digital transformation’ will soon become obsolete, as enterprises achieve full integration. Systems will connect to form an AI-driven automation landscape, and we may not even realise we’re using AI.
Regarding generative models, I expect a significant leap in quality, reduced hallucinations, and better responses to nuanced queries. Imagine a workplace where a person receives an email, and AI drafts a reply mimicking their style — the person simply reviews and sends it. This is something we do every day, and AI could make it much more efficient. This same approach applies to automatic document completion as well. AI-driven automation of manual tasks — what’s now termed hyperautomation — can be implemented easily, delivering cost savings and productivity gains.
One opportunity I’m especially passionate about is AI’s role in driving inclusion and diversity, like enabling real-time translation between spoken and sign languages. In an ideal AI-driven world, decision-makers will interact with machines via natural language, accessing personalised, explainable dashboards 24/7.
Today, we rely on static dashboards and PowerPoint exports — but in the near future, leaders will request insights via natural language, receive tailored visuals, and ask ‘Why is this data here?’ And get the answer. Many companies are already exploring this, and I hope it becomes standard within 2–3 years.
To sum up, companies that approach AI with speed, open-mindedness, and a human-in-the-loop mindset will succeed. The biggest risk is not using AI at all — or not engaging users and learning from feedback. These systems will integrate naturally into operations. In healthcare, AI could help improve the frequency and reliability of second opinions, offering insights and coaching 24 hours a day. That’s a future worth building. One more point that’s critical to acknowledge: recent studies show that only up to 20% of AI projects are currently successful, which is really sad, to be honest. I hope these results will shift dramatically in the next two to three years. Fingers crossed for that.
Lea, how can organisations ensure that AI and data-driven decisions are truly improving patient outcomes, not just operational efficiency?
Lea: That’s a really good question because it really puts the patient into the centre — and this is really the most important part. Patients have all different kinds of patient journeys. They’re very individual in terms of needs and preferences. And we are also talking about humans, and human bodies are super complex. Health isn’t only based on genetics, but also on lifestyle choices and environmental factors. So organisations really need to carefully consider which health gap they actually want to address, and how to develop and place a solution in a way that’s meaningful, accepted, but also enjoyed by the relevant stakeholders.
Katarzyna, how would you assess the state of AI platform maturity across the DACH region? Is the landscape evolving fast enough to support strategic AI ambitions?
Katarzyna: That’s a very interesting question. What I have observed over the last 12 or more years that I have been developing end-to-end AI-driven solutions is that the technology is at a very, very mature level. I would even say we can’t keep up with the speed of the technology that is developing.
What is missing, however, are cross-functional teams to properly prioritise business use cases and collaboratively determine which data is important for each use case. Since data is the essential foundation for every use case to remain a bottleneck. So the call to action is clear: create cross-functional teams and align your data and AI strategy with the business strategy.
Lea, what unique challenges do you see in applying AI to one of the world’s most regulated sectors, and how do we overcome them?
Lea: That’s a really, really important question, because adopting AI in healthcare needs to be trustworthy and ethically sound. It also needs to keep the standards of patient and data protection very, very high.
So AI solutions need to be continuously reviewed by users and be adopted by the feedback which is received, and also be in line with the legal environment and policy infrastructure. We, as AstraZeneca, are currently trying to overcome data accessibility, and we are working on possible solutions. For example, using AI to generate synthetic data sets so that we get a better understanding of data. We are addressing this one pillar for now.
Katarzyna, what’s one belief about AI adoption that you think more leaders need to rethink?
Katarzyna: Thank you — this is a question close to my heart and one of the most important for me. AI is not only about developing technology. It’s very interesting and actually a bit sad to observe that we’re facing the same issues in AI that we saw when the first scaled software solutions were developed.
Data science and AI-driven solutions are not for the product development teams — they are for users. That’s why it’s so important to have cross-functional, diverse teams that include user representatives. I learned this from an outstanding leader I had the pleasure to work with, who said, ‘Who am I to decide how the feature should work? Let’s ask the users.’ That stuck with me for years.
The more we integrate users into the development process, the more user-centric — and ultimately more successful — the solution will be. But it doesn’t stop there. Solutions must be supported by leadership. They have to prioritise it. It can’t be something users are just expected to learn on top of everything else. They need time and support to adopt it step by step, otherwise, it’s just another extra task.
I once worked with about 30 users to understand their pain points and co-create a solution. Initially, they didn’t believe we could help. But after just a few days of embedded work, following them, understanding their challenges, they became so engaged that by the time I got to the airport, the solution design was already in my inbox. The development team was pushed into production within two months because the users were pulling us so strongly. They used it every day and improved it with us step by step.
To sum up, we need to build user-centric, cross-functional solutions, prioritised by leadership, and designed in close partnership with the people who’ll actually use them.
Lea, from your perspective, what skills or collaborations are most essential to make AI in healthcare not just possible, but sustainable?
Lea: It’s really all about communication. So, continuous communication between the healthcare professionals, between the users and also between the AI expertise. It’s really essential to have easy-to-use and explainable AI as well. And it, of course, needs leadership and courage to think of AI, develop AI and really get it into practice.
Katarzyna, how do you personally stay grounded and strategic in a field that’s changing faster than most can keep up with?
Katarzyna: Yeah, it’s very hard. I can tell you — it’s very hard to follow all the technological developments. The speed on the market is incredible. It’s a tough one, but we need to deal with it. First of all, it’s important to listen — listen to users, listen to experts. Read, understand, and reflect. For me, coding is still very important. Of course, I don’t code every day in my current role, but I try at least on weekends to code something, attend a hackathon, or stay close to the code.
I started my tech journey in competitive intelligence, trend development, and market research. I still use those skills daily — just with different models and tools. To keep it short: I use AI to stay updated about AI. And I use my deep knowledge to check for hallucinations. When I read a text, I need to verify if it’s true or not, because we know that AI models can still hallucinate. They’re improving, but we still need to be careful. That’s where the human-in-the-loop comes in.
Additionally, I attend conferences like Tech Show Frankfurt, after-work events, and connect with peers. I follow my role models — like Cassie Kozyrkov — on LinkedIn. And I write. As we know, when we write, we explain the content to others and we learn even more.
Lea and Katarzyna, if the audience walks away from your session remembering one idea, what would you want that to be?
Katarzyna: How to unlock the full potential of AI in pharma.
Lea: I absolutely agree.
– – – – – –
Together, Katarzyna and Lea bring both strategic vision and hands-on expertise to the challenge of scaling AI in one of the most complex and highly regulated industries. From platform maturity to public health impact, their work highlights how collaboration, curiosity, and a shared commitment to real-world value are shaping the future of data-driven healthcare.
Catch their joint session “Unlock the Full Potential of AI in Pharma” at Big Data & AI World Frankfurt, part of Tech Show Frankfurt, on 4 June 2025 from 13:40–14:10 at the Keynote & Strategy Stage.
Editor’s note: All quotes are verbatim from a recorded interview, lightly edited for clarity where necessary.
