The Evolution and Impact of AI Tools in Medical Writing

Introduction

Artificial intelligence (AI) has become engrained in our everyday lives and, broadly speaking, we seem accepting of its integration into our smart devices, cars, and consumer electronics where it promises to improve user experience, save time, or reduce the burden of mundane tasks.

In the data-driven world of clinical research, it is not surprising that companies, especially larger pharmaceutical companies with multiple programs and all the associated documentation, view AI as an opportunity to enhance productivity. So, what does this mean for medical writers, and is the wider adoption of AI into our working practices something to be worried about or something to be embraced?

Spoiler alert, I can’t answer that for you. Truth be told, I’m still trying to work it out for myself.

What I’m sure we can all agree on, is that AI is certainly coming our way, and there was no clearer indication of this than at this year’s European Medical Writers Association (EMWA) Spring conference in Valencia, where the best part of 2 days was devoted to talks on AI from a range of different stakeholders, and the topic of AI dominated many breakout discussions.

AI in Medical Writing: Basic Tools vs the Future

Software tools to aid the authoring of documents are nothing new (anyone else remember Clippy, the office assistant?). Medical writers already benefit greatly from the tools and macros packed into Microsoft applications, and third-party add-ons, that will check our spelling and grammar, automatically number our tables, and create our tables of contents and reference lists, to name a few. Most of these tools are based on simple algorithms. We’re comfortable with these, and if we’re honest, we couldn’t live without them nowadays. While some software packages, like PerfectIT, go a stage further and are based on advanced algorithms with some AI-like features, others, such as Grammarly, are considered true AI tools due their natural language processing and machine learning capabilities. So, AI is already in the mix, and able to help us with basic functions like maintaining consistency, text suggestions, and adhering to language and style rules; should there be any concern if AI encroaches further?

With the basic tools currently at our disposal, we are firmly in the driving seat as the content creators, data interpreters, and decision makers, but with AI theoretically able to perform a lot of those ‘human’ tasks, it is understandable that medical writers might be feeling a bit nervous.

I think most current concern exists because we don’t know what the AI tools will ultimately look like, what they will do for us, and what they might take away from us as medical writers. Where will that line between human and ‘machine’ be drawn?

Potential Benefits of AI in Medical Writing

The main draw of AI for larger pharmaceutical companies, is that AI solutions could enable medical writers to handle larger volumes of work, making it possible to scale operations without compromising on quality.

Reducing the time required to produce medical documents by leveraging AI for activities like data extraction, generation of descriptive summary text, and maintenance of terminology and style consistency within and across documents could allow writers to focus on more strategic tasks. The advanced analytics provided by AI tools could also offer deeper insights into clinical data, thereby allowing writers to craft more informed and data-rich documents.

Challenges and Considerations of AI in Medical Writing

While the potential benefits of AI for medical writers are notable, integrating AI into medical writing comes with several key challenges:

• Ensuring the quality and reliability of AI-generated content remains a primary concern, so continuous oversight and validation by human experts will remain necessary to maintain standards for the foreseeable future. Medical writers will therefore need to be train in the effective use of AI tools, making investment in learning and adapting to these new technologies essential.
• Handling sensitive patient data requires stringent adherence to privacy regulations. AI tools must be compliant with laws like GDPR and HIPAA to prevent data breaches. Further to this, if we want to truly embrace the machine learning capabilities of AI for medical writing, the tools need lots of data to learn from. How will this be achieved when patient and client confidentiality, and protection of intellectual property, might limit the availability of data?
• The use of AI in generating medical documents also raises ethical questions about authorship and accountability. Clear guidelines and policies must be established to address these issues.

As the director of an agency that employs medical writers and provides writing services to a range of clients, large and small, I foresee a few additional challenges too:

• Some larger pharmaceutical companies are already developing their own bespoke AI tools, either in-house or in collaboration with AI developers. Providing medical writing assistance to these companies in the future will probably mean accessing and using their AI tools, like we do with the other systems they use, but what about smaller pharmaceutical and biotechnology companies without the means to develop their own? Will there be off the shelf AI tools available for them, considering they won’t necessarily have the standardisation (templates, outputs, etc.) and breadth of data to train systems on? Will there be tools available to agencies that we can use as part of our service offering?
• Developing and validating AI tools for use in our tightly regulated industry will be expensive, so what will the cost of AI tools be for agencies like ours and smaller pharmaceutical companies – will it be financially viable to employ them?
• Which AI tools should we be looking at? I think it’s fair to say it’s still early days, and while several companies are developing AI tools, they are mostly at the beta testing stage, and will likely take time to come to market. Depending on progress and competition, some might not make it to market.
• As a medical writer that started at the bottom with data checking (QC) tasks, manually trawling through outputs to populate in-text tables and digging deep into listings to learn, I do worry that newer writers may not develop the breath of understanding and experience with certain documents that comes from those ‘entry level’ tasks if they are done for us. Further to this, is there a risk that people will devalue the role of medical writers if they come to believe AI is doing the hard work for us?

The Future of AI in Medical Writing

I’m conscious when I scan up the page that the section I’ve devoted to the potential benefits of AI is considerably shorter than the section I have on challenges! This is in no way an indication that I’m against AI tools for medical writing; it simply demonstrates that I have many questions, currently, and I know I’m not alone.

It is tempting to think that at some point in the future, the generation of a clinical study report (CSR) could be entirely handled by AI, at the push of a button, thereby making the medical writer obsolete; but is this likely? From talking with peers and the AI developers tending their stands at the recent EMWA conference, I think this is unlikely. AI tools will still need medical writers to ‘drive’ them and then tailor and check the output they deliver. It is also likely that the role of AI in the writing of (for instance) a CSR will be largely confined to helping the development of the first draft, with the finessing of subsequent drafts falling to the medical writer.

Conclusion

In summary, AI tools are slowly but surely being integrated into various medical writing processes. Whilst many tools are still in beta testing, the aim of integrating AI tool into medical writing is to efficiently deliver accurate and consistent documents. While there are challenges to address, the potential benefits should be welcomed by medical writers. Embracing these technologies will be key to staying ahead in the ever-evolving healthcare communications landscape. As AI continues to evolve, we anticipate that its role in medical writing will expand, paving the way for innovations that will further enhance the quality and impact of healthcare communications.