AI-enabled automation of Safety & Regulatory process: where are we now?

Lucy Walters, PharmiWeb (LW): What stood out most for you in the ArisGlobal 2024 Industry Survey Report[1] findings?

Emmanuel Belabe, ArisGlobal (EB): The gap between use and intention. Although 75%+ of surveyed organisations claim they already use some form of ‘advanced automation’ within their processes today - up 13% from 2022 and just 5% in 2020, only 8% have applied next-generation capabilities in “all” or “most” of their processes at this point. (By advanced automation, we mean the adoption of artificial intelligence (AI) and machine learning (ML).)

As AI and ‘deep learning’ solutions advance in line with the scale and sophistication of available data, life sciences R&D organisations are growing more ambitious in their ability to harness data and its insights in ever smarter ways. That could be to hone decision-making, drive efficiency gains, and deliver important treatments to patients more affordably.

It was the pace of companies’ transitions to intelligent automation that we were particularly keen to identify in this latest study, and we found that 60% of organisational leaders were looking to explore new usage and/or increase usage of AI/ML over the next 18 months.

LW: What has prompted this?

EB: Typically respondents see automation and AI as having the potential to increase positive outcomes and empower professionals across the drug development cycle. Increased adoption of these technologies is expected to benefit patients, through increased drug discovery, personalised therapies, clinical trial acceleration, enhanced care, and improved data analysis. Other anticipated business benefits include increased speed, superior decision-making, and more efficient processes – all critical success factors in a highly competitive industry.

LW: What about challenges, or barriers to adoption of these more advanced capabilities?

EB: Many organisations’ next-level AI ambitions are inhibited by integration challenges. While the most substantial blocker to AI adoption in life sciences is budget (cited by 53.6%), a notable 36% highlight lack of integration with existing technology. More than two-thirds (68%) of respondents went on to say they considered or had found it “very difficult” or “somewhat difficult” to integrate automation technology with other systems and/or data.

This barrier to advanced automation goes some way to explaining some of the identified hesitancy around where the technology could take companies in the future. That’s particularly as new opportunities emerge to exploit real-world data (RWD) as part of critical but labour-intensive processes such as safety signal detection and validation.

LW: What did the survey uncover in terms of companies’ plans for more ambitious use of data, across safety and regulatory operations?

EB: There is an appetite to harness RWD where possible to unlock the next level of patient treatment innovation. In our research, just over half (51.2%) of respondents suggested that their organisation is already connected to some form of RWD, up from 31.5% in 2022; moreover 54% of those currently connected to RWD said they are looking to increase the data’s usage.

For those not yet connected to RWD, 20.4% indicated that they are either in the process of connecting to such resources, or have plans to within the next 18 months.

Access to high volumes of good-quality, current data will be important to build Large Language Models to drive Generative AI’s use in the industry, too. As it was for early automation, data is the key driver of Generative AI. The more data an organisation can leverage, the stronger GenAI’s models (algorithms) can be accurately trained to help identify and qualify the next raft of industry breakthroughs.

According to respondents, the most important value from RWD in drug development is linked to clinical and patient safety activities. Respondents indicated that the ability to draw on and analyse additional data and insights could help de-risk clinical trials, expedite the trial process, speed up approval processes, and positively impact signal detection.

LW: Was there anything else of note to come out of the study?

EB: Yes - that regulatory bodies/health authorities will be crucial in the success of advanced automation in promoting safe new innovation in the industry. Without careful management, approaches, and implementation, RWD data use could become more of a point of contention in interactions between regulatory agencies and organisations, a number of respondents indicated. RWD is being used within the US FDA’s Sentinel Initiative and by EMA in Europe, but there needs to be further partnering to pave the way to harnessing RWD in other use cases. 

The full industry report that explores this topic in more depth, AI & Automation: How does your company’s progress rank compared to the findings of our new industry survey?, is available for download here.

[1] ArisGlobal 2024 Industry Survey Report: Life Sciences R&D Transformation: Ambitions for Intelligent Automation & Today’s Reality