Augmenting the Regulatory Worker: Are We Making Them Better or Worse?

Artificial intelligence vs human

In the precipice of generative artificial intelligence (GenAI), discussions surrounding how humans use AI tools are becoming more and more frequent in the industry. In addition to AI’s existing challenges surrounding discrimination and data accuracy, GenAI presents a new problem: hallucinations. These hallucinations create realistic-seeming content not based in reality. Because GenAI is only as good as the data it is trained on, high-quality data and human oversight are imperative to avoid passing off false information as fact. With as many as 77% of biotech and pharmaceutical companies stating that they were monitoring the latest technologies for use in regulatory processes (Why Biotech and Pharma Companies Are Embracing Regulatory Technology Outsourcing), conquering these challenges becomes especially important for successfully augmenting regulatory workers.

Michelle Gyzen, senior director, regulatory affairs and drug development solutions at IQVIA, stated, “There are major concerns coming up in client and internal discussions focused on ‘are we making regulatory workers better or worse by augmenting them’ especially now that the industry is facing adopting GenAI.” She further explained that while there has been an enormous amount of data collection and regulatory standardization completed, the industry can still do a great deal more.

Michelle Gyzen [square]
Michelle Gyzen, 
Sr. Director, Regulatory Affairs
& Drug Development Solutions

The solution is not a plug-and-play module for the industry. Classification models need to be rewritten and data schemas evaluated for accuracy, but most importantly data scrubbing to ensure clean is of utmost importance because the data cannot be trusted otherwise. Addressing these challenges is the first step to the successful use of GenAI. Given how fast AI is evolving and being adopted, addressing the clean data challenges is imperative simply because “we still don’t know the full scope of capabilities in terms of what AI, and especially advanced AI, can do,” Gyzen said.

Clean Data

Because AI learns from historical data, which is unfortunately often discriminatory and biased, there is an inherent unconscious bias lurking in AI. The industry is grappling not only with initial bias, but bias creeping into what was once clean data.

Since GenAI is in its early stages, Gyzen recommends a phased approach to implementation. Gyzen is working internally and externally to ensure data accuracy, interoperability, connecting with various regulatory information management (RIM) systems, and integrating regulatory intelligence with current AI tools. All of these steps must be completed before venturing into GenAI. In her opinion, stringent processes such as this are the only way forward for the industry, especially for regulatory processes. “As we evolve with AI, we need multiple layers of validation for regulatory intelligence data in place. This can come in the form of using an automation program to understand the data first and provide summary synopses. But here is the key element: having true human oversight viewing that data not only from a global regulatory perspective, but from a local regulatory intelligence perspective as well.” Gyzen echoed the sentiments of other thought leaders present at the 2023 BiotechX USA when she notes, “while we are close, the technology is not there yet to take the training wheels off.”

The Human Factor

Another factor to keep in mind during AI adoption is the human, in this case regulatory workers. Leading with technology for AI adoption is fundamentally flawed. “The idea that companies are going to be able to put technology in place and everyone is going change their processes and the way they work is not realistic, so technology needs to be built for humans and the way humans work,” Gyzen said. This understanding of regulatory processes and human behavior is what companies need to keep in mind to outline what can be augmented and build processes that enable rather than hinder regulatory professionals.

Industry Data and Pulse

With half of the industry outsourcing (54%) and the other half building in-house (46%), the industry is clearly split on the best approach for AI technology (Why Biotech and Pharma Companies Are Embracing Regulatory Technology Outsourcing). Although many larger biotechs and pharmaceutical companies have the resources to build their own in-house, they are not necessarily looking to do so. In a live panelist discussion with representatives from Amgen and Allogene Therapeutics at BioTechX USA, it was stated that they are not necessarily looking for in-house production, they just want something that is cost-effective and works. Gyzen agrees, “With larger organizations, there’s opportunity to partner because they don’t necessarily have access, to nor the understanding of, what the entire industry is working on. This is the key value that vendors and outsourcing partners bring: just having a greater understanding of the pulse of the industry.” Gyzen finds that smaller organizations benefit from this knowledge as well, but the real value is access to larger data pools and safeguards to protect customer information and data.

At present companies are looking for full integration. This is something that Gyzen routinely comes across in her role. “A big part of my role here is interoperability in developing regulatory systems and what that means from a services standpoint. Because IQVIA is partly a services organization, we have multiple customers with multiple types of infrastructure. What I’m looking at is ensuring the ability to integrate not only cross-functionally and across multiple systems, but cross-organizationally.” As technology evolves, Gyzen highlighted that she focuses on integrating with client systems to create a seamless flow of data through multiple enterprises and organizations across the globe. This is a very challenging task because closed-box systems create a technological bubble.

For companies to successfully evolve as AI moves to GenAI, clean data needs to flow into the appropriate systems. Currently, challenges surrounding not only clean, accurate data, but data access can hinder biotech and pharma companies. As Gyzen states, “We live now in a world where the market is so very segmented and so very fragmented that you know we have to come up with a solution if we’re going to move forward.”

The insights team analyzes and comments on industry trends and creates thought leadership content for BioSpace and clients. The head of insights, Lori Ellis, can be contacted via Follow her on LinkedIn.

Back to news