Artificial intelligence (AI) is making its way into every sphere of the modern way of working, helping us to collect and interpret vast amounts of information and transforming the global infrastructure of evidence generation, synthesis, extrapolation and interpretation. It seems clear than when it comes to the use of AI when dealing with evidence, traditional methods cannot win. Does the old proverb “if you can’t beat them, join them” apply here? Are we facing a takeover or a merger? Let’s discuss.
AI algorithms are being harnessed in evidence generation due to their ability to organise, process, transform, and analyse large data sets, transforming unstructured data into structured, useable formats. “Automate your data aggregation and focus on what really matters…” These types of statements now regularly pop up on our LinkedIn pages and advertise the opportunity to integrate AI into our evidence generation activities. AI is being sold to researchers as a silver bullet, a valuable, time-saving tool to alleviate the cumbersome tasks of collecting and processing healthcare data.
AI integration was one of the key focus points at The Professional Society for Health Economics and Outcomes Research (ISPOR)’s recent conference in Barcelona. Specifically, ISPOR attendees learned that machine learning and natural language processing approaches have the capacity to quickly and optimally aggregate unstructured clinical data e.g., electronic health records, and to generate real-world evidence data that can be leveraged in product development and market access activities. These powerful algorithms also demonstrate the utility of AI-powered analytics, which can be used to identify trends, unlock relevant correlations, and evaluate the long-term effects of a treatment. These abilities find their relevance in lengthy and costly medical activities such as precision medicine, drug development, and scientific research. The future of medical research may be greatly impacted by the adoption of AI-driven tools.
Evidence synthesis has not been untouched by the rise of AI, either. Some have stated that the application of AI in synthesising evidence will result in better results for less money. Systematic literature reviews (SLRs) are an incredibly resource-intensive process, requiring the identification, analysis, and summary of all available evidence in response to a review question. Machine learning tools can be used in in these processes to reduce the need for repetitive tasks and improve efficiency.
Concerning their use in writing and SLRs, the Cochrane Collaboration mandates that:
- AI tools cannot be credited as authors
- Authors bear full responsibility for the article’s accuracy and validity
- A transparent and detailed description of the AI tools used, and content generated is mandatory
The establishment of these rules does not appear to have inspired a ‘full steam ahead’ approach to the use of AI in SLRs. Despite the go-ahead to utilise AI in evidence synthesis so long as authors remain transparent, the industry seems to have been slow to uptake. Some have surmised that its slow uptake is due to “the field having grown to equate human effort with methodological quality, such that automation may be seen as sacrificing quality.” Is this really the case? Has the emphasis on a slow and steady approach concealed an outdated perspective within the industry regarding what constitutes high-quality evidence? Let’s revisit the fount of rules and regulations in evidence synthesis, the Cochrane Collaboration, to understand their attitude about the adoption of AI in producing high-quality health evidence.
In May 2024, Cochrane released a statement on their website stating they were collaborating with experts to plan a way forward for Cochrane to benefit from AI. This plan will include best practice standards for its use, the requirement for current tools to be validated with a well-defined pathway for endorsement, an established robust foundation for tool implementation, and research that investigates the use of AI tools in medical research. The tone from Cochrane was hopeful, rather than antagonistic. Currently, the final draft of the consensus-based Responsible use of AI in Evidence Synthesis (RAISE) guidance and recommendations is being finalised. Cochrane has scheduled a series of webinars and training in 2025 leading up to its publishing, fuelling the buzz around AI and its incorporation into the market access and evidence synthesis space.
Similarly, in August 2024, the National Institute for Health and Care Excellence (NICE) published their position statement on AI, although this came with a degree of ambiguity. NICE recognised the potential for AI to enhance the HTA process by supporting evidence generation and synthesis through leveraging the predictive capabilities of machine learning algorithms, acknowledged the increasing role of AI in the market access space, and emphasised the need for caution by stating that AI “should only be used when there is demonstrable value from doing so.” For companies considering the use of AI in evidence generation, it is advised they seek NICE advice through early engagement, and in later stages of evidence development, that they discuss their plans with appropriate NICE technical teams.
At Initiate, we believe that AI may allow us to increase human capacity in evidence generation and synthesis. But not quite yet. While we may be steadfast in our approach, we are committed to adhering to the most rigorous methodologies in our work and so before we venture into the unknown, we want to be armed with the best practices from the RAISE guidance. Rest assured, we are closely monitoring developments and guidelines ensuring we stay up to date with changes in the industry. In 2025/26, we will be ready to join the line to cautiously introduce AI algorithms into our evidence synthesis practices, while maintaining a watchful human presence throughout the process. In the meantime, we will continue to assess the potential utility of these tools and explore potential weaknesses and bias that their inclusion may introduce.
At Initiate, we pride ourselves in being ready and prepared to help guide our clients through changing times. We have experience in all major global markets and expertise across a wide range of disease areas. If you would like to discuss any of the developments outlined above, or any other market access or HEOR needs, get in touch at hello@initiateconsultancy.com.
