Visit The Newsroom | Insights from Subject Matter Experts

AI and Medical Communications Development: Striking a Balance

Written by Alphanumeric | Sep 30, 2024 5:46:00 PM

 

Presented @ Digital Pharma East - Sept 11, 2024

The integration of AI tools into medical communications has sparked both excitement and caution across the healthcare and pharmaceutical industries. While AI promises efficiency, cost savings, and innovation, striking the right balance between automation and human oversight is key. In a recent discussion at Digital Pharma East 2024 led by Dr. Christina Nixon, Senior Director of Medical Communications at Alphanumeric, and Dr. Preeti Sule, a leader in scientific communication at Moderna, the duo explored how AI is being incorporated into medical communications. Together, they reviewed three practical use cases for AI in medical communications and the steps required to ensure successful implementation.

Baby Steps: Leveraging AI for Literature Summaries
One of the first ways in which AI can be utilized in medical communications is through generating literature summaries. Dr. Nixon explained that starting with literature summaries is a manageable introduction to AI. In this process, a pharma partner or agency can generate a draft using AI tools, like GPT models, which can then be aligned with strategic messaging, data validation, and quality control by human teams before being reviewed by the pharma partner for final approval.

Dr. Sule emphasized that while AI can speed up the process, the real challenge is ensuring that the summaries remain closely tied to the product narrative. As she noted, new data arises and company priorities and narratives can shift rapidly, making it crucial for AI-generated content to be continually refined to fit the evolving landscape.

Roadblocks: Despite its potential, early efforts to generate literature summaries with AI faced significant challenges. The most prominent hurdle was aligning AI outputs with the company’s messaging. Dr. Sule illustrated this with a clear example: initial AI-generated summaries missed the primary points of the company’s message, failing to highlight key issues such as improved efficacy for specific patient groups.

To overcome these issues, the panelists stressed the need for "human crawlers" to review the AI-generated content and ensure that essential messages were not missed. While AI can handle some of the heavy lifting, human oversight remains essential for quality control, even in a simple output like a literature summary.

AI for Speaker Guides: Balancing AI and Human Expertise
Next Dr. Nixon and Dr. Sule discussed using generative AI for developing speaker guides for Key Opinion Leader (KOL) interviews. Compiling questions and answers into a speaker guide for video interviews can be a labor-intensive process. AI can simplify this process by generating possible responses and appropriate references for an initial draft, which can then be aligned with product messaging saving time and resources. AI can also facilitate the production of post interview transcripts, streamlining the documentation and approval process.

Roadblocks: Even with the promise of efficiency, the process is not without its challenges. Dr. Sule noted that, despite AI's ability to generate useful content, AI tools still struggle to handle the nuanced requirements of KOL guides. One of the major issues was ensuring that the  outputs were not only accurate but also tailored to the specific target audience. For example, AI-generated responses for pharmacists must be distinct from those created for pediatricians or other healthcare professionals. Failing to properly segment these audiences’ risks delivering messages that lack relevance and impact.

The importance of prompt engineering—the art of crafting precise instructions to guide AI tools—was another critical takeaway from this discussion. As Dr. Nixon pointed out, the way a question is framed significantly affects the quality of the AI's response. Experimenting with different prompts is essential to generating useful content.

AI-Generated Voiceovers: Simplifying Post-Production
A particularly exciting use of AI in medical communications is in the creation of AI-generated voiceovers for video content. When pharma companies produce video material, regulatory and compliance committees often request changes, which can require re-recording parts of the audio. AI-generated voiceovers offer a streamlined solution to this issue. Once the script is approved, an AI tool can generate audio in different languages, tones, and styles, allowing for rapid corrections and edits without the need to bring back voice actors.

Dr. Nixon demonstrated how quickly and easily voiceovers can be generated and modified with AI. Within just a few minutes, she was able to produce three different versions of the same audio file, showcasing the flexibility and speed of AI in adapting content to different tones and styles.

Roadblocks: The ease and speed of AI-generated audio are tempered by some important limitations. One major challenge is maintaining consistency and clarity across different languages and audiences. As Dr. Sule pointed out, AI tools must be programmed to account for tone, pronunciation, and linguistic differences, which can vary significantly between regions. These subtle differences must be carefully managed to ensure that the message resonates appropriately with each target audience.

Additionally, small changes to audio files can cause headaches during the video editing process. Altering a phrase in the audio track may lead to more complex edits if the video production team needs to realign the entire audio file with the visual content.

The Importance of Human Oversight and Training
Throughout the discussion, one key theme emerged: while AI can augment the work of medical communicators, it cannot replace human expertise. In every case study presented, from literature summaries to speaker guides to voiceovers, human intervention was crucial for ensuring that AI tools delivered the right outcomes.

In particular, Dr. Nixon and Dr. Sule emphasized the need for prompt engineering skills within teams that work with AI. To get the best results from AI, teams need to know how to craft specific, well-thought-out instructions, ensuring that the tool delivers content that aligns with both the scientific data and the company’s strategic narrative.

As AI becomes more integrated into medical communications, companies will need to invest in training their teams to work effectively with these tools. Whether it's using AI for brainstorming, drafting, or producing content, the best results come from teams that understand how to guide AI’s capabilities toward their desired outcomes.

Looking Forward: AI and the Future of Medical Communications
While AI’s role in medical communications is still evolving, Dr. Nixon and Dr. Sule’s discussion highlighted both the potential and the challenges that lie ahead. AI tools can streamline processes, reduce costs, and help pharma companies produce content faster, but the technology still requires human intervention to ensure accuracy and alignment with company goals.

As AI continues to develop, its ability to generate high-quality content will improve, particularly as companies invest in more advanced models like enterprise GPTs that can be trained on specific industry lexicons and narratives. However, the need for human oversight will remain critical, especially when dealing with the nuanced and highly regulated world of medical communications.

The future of AI in medical communications is bright, but it requires a careful balance between automation and human expertise. By taking small steps and learning from each experience, companies can unlock the full potential of AI while ensuring that their communications remain accurate, compliant, and impactful.