AI in biopharma marketing: Balancing opportunities and risks

Biopharma companies are beginning to capitalize on generative AI (GenAI) to accelerate the production of highly personalized, localized digital content for healthcare providers (HCPs) and patients. But to make the most of GenAI, their teams must successfully address several key challenges. In particular, they need to comply with data privacy regulations while also protecting sensitive company data.

In our webinar, a panel of AI, legal, and digital experience experts discussed the opportunities and challenges of integrating AI into biopharma marketing. The panelists agreed that implementing some best practices can help teams minimize risks and maximize the value of AI to deliver accurate, relevant, and compliant content at scale.

AI can help biopharma companies provide accurate and relevant content directly to HCPs and patients. Pharmaceutical manufacturer BERLIN-CHEMIE, for example, decided to integrate AI capabilities into its digital health portal. The company wanted to deliver personalized information to its patients through a chatbot, called “ISA.”

The launch of a new 12-week program on the portal, which focused on chronic obstructive pulmonary disease (COPD), offered an opportunity to create an interactive experience that could present patients with relevant information about the diagnosis and its implications. With help from IBM iX and Magnolia, the BERLIN-CHEMIE team incorporated the chatbot into its portal to deliver that information.

“The normal healthcare consultation time is about ten minutes—that is very little time for patients to process a new diagnosis,” says Jan Hulshoff Pol, head of digital healthcare marketing at BERLIN-CHEMIE. “Patients often go to Google and start searching. And what they get is mostly SEO-optimized content. ISA provides certified content through the platform to better support patients.”

To find the best information, ISA draws from over 700 scientific articles on COPD—more information, and better information, than patients could find on their own. ISA then presents content that is most relevant for each individual patient.

Whether biopharmas are using AI to enhance patient health portals, provide drug information to HCPs, or support another use case, they need to ensure that all content adheres to strict regulations. Providing inaccurate information could put patients’ health at risk and expose a company to serious liability.

At the same time, biopharmas have to protect their intellectual property. Organizations must make sure that information cannot leak out through a GenAI app, for example. Moreover, biopharmas need to be sure they are not using another organization’s protected information, their own trade secrets, or a patient’s personal information to train a model.

The recently enacted EU AI Act intends to clarify which AI applications and systems are legally regulated, outright banned, or left unregulated. According to that act, several biopharma uses of AI could fall into “high-risk” categories that are tightly regulated. “The legal requirements depend on the use case - what you’re planning to do with AI,” says Boris Arendt, attorney Simpliant legal. “It makes a difference if you use the AI for your own purposes, like marketing, or if you want to implement AI into products, like a medtech device.”

Working with a legal expert to evaluate AI use cases is critical for avoiding regulatory violations, especially as new regulations and rules emerge. “The legal landscape is still evolving,” says Arendt.

Just as humans should be involved in regulatory reviews, they should be integrated into AI-aided content workflows. GenAI tools, for example, require humans to review the quality and accuracy of generated content, helping to identify any potential errors or AI hallucinations. This is particularly important for biopharma companies, since errors could have serious implications for patient health.

Humans can also optimize content for its intended audience, ensuring it has the right structure and tone of voice. “Last year, we were all focused on prompt engineering,” says Christina Schiffler, executive director and design principal at IBM iX Studio Berlin. “A prompt engineer is important. But we also need content designers who can understand how to structure content and deliver a result that is good and relevant.”

The good news for many biopharma marketing teams is that they can start exploring and benefiting from AI rapidly. “The pure technical hurdles are relatively low compared with other types of projects,” says Jan Schulte, head of group consulting at Magnolia. “Because in the end, what we are speaking about is just making good use of APIs.”

Employing the Magnolia AI Accelerator, for example, marketing teams can easily tap into multiple GenAI engines within a single, unified interface to start creating personalized content quickly. They can use Magnolia Hyper Prompt—a tool within the AI Accelerator—to optimize prompts, ensuring that results are on brand and follow regulatory guidelines.

The bigger challenge is focusing core data on a specific domain, such as biopharma. If biopharmas draw on too much data from beyond their domain, they risk generating incorrect or irrelevant answers. Carefully curating data is key. “Once you have all the data that really matters, the reasoning capabilities of modern LLMs are really amazing,” says Schulte.

Ready to learn more?

For more insights from this expert panel, watch the webinar, “AI in biopharma marketing.”

To learn how to prompt like a pro, watch the step-by-step walkthrough with Magnolia expert Jan Schulte in the webinar “Bringing AI to content workflows.”

For additional exploration on GenAI use cases and key implementation decisions, download the white paper from Magnolia and IBM iX, “Getting started with GenAI in biopharma marketing.”