In this post we continue our series on AI for Medical Affairs. Previously, we covered how AI can turbocharge the strategic planning process. Here, we tackle another common Medical Affairs task – developing a scientific communications platform.
A Scientific What Now?
A scientific communication platform is a series of scientific statements that tell the scientific story of a product in a given therapy area, presented alongside the supporting references. Usually initiated well before regulatory approval, the scientific platform is an important asset in ensuring accurate and consistent representation of the educational story. Key assets within the scientific platform include the scientific story, scientific statements, and high-level evidence summaries – all supported by published evidence from the scientific literature.
Let’s look at how most Medical Affairs teams go about developing the platform…well it all starts with the scientific story. With so many drugs out there and so many different treatment options, it is important to define the potential role this new drug or approach has in practice. This story is then supported by the scientific statements – ‘key phrases’ on a particular topic. Ultimately, when the data become available, these statements should be linked to highlight a given result. These scientific statements must respect every aspect of the science of drug development and provide data-led, fully—referenced (and thereby supportable) context to how any new therapy fits into the current landscape. Without these considerations being front of mind, a scientific platform is simply an unsubstantiated wish list.
Now this is where developing a scientific platform becomes interesting but time-consuming. Given the complexity of many aspects of medicine, there are usually a good number of potential scientific statements. Developing a scientific platform means finding, reviewing, assessing, and then validating the publications on basic science through to the clinical studies and real-world evidence. To be useful and useable, the scientific statements must be robust and defendable, and reflect all the relevant literature.
After the scientific statements, comes the scientific summary – this rolls the key results into a high-level overviews providing that all-important context. As the volume of data increase, there is also merit in aggregating results to make the overall findings more accessible to the reader.
Once completed, a scientific communications platform is a powerful internal tool to support a broad range of activities across the entire organization:
- Medical affairs, commercial and health economics and outcomes research (HEOR) teams: to curate and maintain the core evidence bank to support content creation across the cross-functional team
- Medical affairs: to identify and substantiate the educational needs and gaps for medical education activities, including publication plan development
- Medical affairs/medical information: to develop and update standard response documents as data become available
- Clinical development and evidence generation research teams: to identify evidence gaps, support protocol development, understand the endpoints, and those all-important patient reported outcomes and real-world evidence needs
So how can AI help?
At its core, the scientific communications platform is about aggregating, curating and then updating the published evidence to support consistent and accurate medical communications. This is an ideal situation for the machine to focus on data gathering (evidence) and curation and the human to focus on the creative aspects (insights, context and communication).
Let’s step back for a moment. PubMed – our main source of biomedical literature – has ~30 million citations and is growing fast. From PubMed’s own statistics, in 2018 alone, approximately 900,000 citations were added compared with ~813,000 in 2017, and the year-on-year trend for substantial growth looks likely to continue.
First and foremost, the scientific communications platform is based on published evidence. It’s a scientific communications platform, after all. As we’ve said before, AI can help uncover the evidence that humans can then review and assess – forming the foundation of the scientific communications platform.
Therefore, AI can help pull out possibly relevant data that are available on a particular topic – this shapes the direction that any scientific statement can take. Where AI comes into its own is that it isn’t biased – it simply reads the abstract extracting the key phrases and brings everything together in a structured manner. Why is this important?
Well, this means that AI considers all the literature available and isn’t restricted to the search terms the user enters – in some cases this results in pulling up 30% more abstracts than manual searches. So here, AI helps to ensure the scientific platform is as robust as possible.
Remember that AI cannot add context or add insights, it is the human who takes all the datasets and collates them to those that are relevant – AI does, however, make it considerably quicker to access these data, draw comparisons, and pull out the key points.
As an example, here’s a short video demonstrating infection rates for nails, supporting a potential scientific statement. (Note: this is meant as an example, not to demonstrate the comprehensive search!) As shown, based on their clinical knowledge, the author writes the content of the statement, and the platform fills in the results table automatically, all based on results that the AI surfaced and which the human had approved. The machine is helping where it helps best, making this process far more efficient.
But…data generation doesn’t stand still – PubMed alone, adds nearly 2500 new biomedical records every day. So, beyond just the evidence, timeliness is an important factor. Communications platforms are a guide and must evolve in tune with the data-led world around it. New results and new competitors may emerge, new findings may highlight issues that weren’t previously as relevant, or cast a new light on established practices. Remembering why scientific platforms are delivered in the first place, and how they are used, new data may warrant minor or major revisions to the scientific statements and summaries – the content cannot remain static. It must update in sync with the world.
In fact, just as AI helps identify the pertinent results in the first place, it can also help keep the platform in sync. The Evid Science platform will alert you of new, related results, and then you (the knowledgeable human) can decide whether or not to include them in your scientific statements. If you do include them, our platform will update the results behind the statement itself!
Here we add a result to our existing statement on nail infections (picked randomly, just to demonstrate), and the table updates automatically, with the reference added. The number of papers in the text changed too, from 6 to 7. (Here, we didn’t update the text of the statement itself, but it would be simple to do so.)
Blending AI & People Provides the Real Value…
To be clear, a scientific communications platform is not simply regurgitating the results from the literature or a database of key findings. It’s an active exercise in summarizing, prioritizing, and contextualizing clinical results to focus communications, identify gaps and data generation requirements, and the strategies to support responsible and robust communications. To do so, anyone developing a scientific platform needs to consider the full weight of biomedical evidence – a task that is both time-consuming and painstaking.
This is the perfect opportunity for machines to do the heavy lifting where they can, leaving people to excel at the people tasks. Gathering the results and data is an ideal machine task (Evid can help with that – the system can process millions of pieces of data in an hour), and the summarization, standardization and prioritization are great people tasks (after all, AI probably couldn’t add the insights or understand your stakeholders’ priorities and weigh them against one another).
This division of labor blends human ingenuity and insight with automation in a deeply complimentary manner. The AI can’t develop your scientific communications platform, but it can certainly help get the data.
So, if you are interested in how the Evid Science platform can help automate some of your scientific communications platform, or how we work with our partners such as the Healthcare Consultancy Group to develop these platforms, drop us a line. We always love to hear from you.
This post was co-written with Rajni Dogra, a communications planner at the Healthcare Consultancy Group (HCG), an Omnicom company.
Rajni has over 20 years of experience in strategic communications and publications activities for ethical pharmaceutical products covering medical affairs, commercial, and access. Her interest in AI stems from an understanding of the importance of representing the scientific literature accurately, no matter who the audience and how fast the turnaround.
Working closely with the Evid team, Rajni and the HCG team are excited to implement AI integration strategies to transform communications activities that are vital in bringing medicines closer to the patients who need them.
In this time of information overload, this team believes that strong collaboration with medical communications experts and data scientists who truly understand medical and clinical information, will yield the greatest opportunities to accelerate adoption of practical AI strategies. If you would like to contact Rajni, please do at firstname.lastname@example.org