A Roadmap for the Use of Artificial Intelligence in Managing Trial Master Files

In a recent video interview with Applied Clinical Trials, Rob Jones, product manager, TMF practice area, Pharmalex, discussed challenges in the trial master file (TMF) space, highlighting outdated regulations such as ICH E6(R3). When it comes to innovation and compliance in the space, he stressed the importance of a risk-based approach, focusing on critical records and leveraging artificial intelligence for big data analysis. Moving forward, Jones believes the TMF community's collaborative nature can be viewed as a strength in navigating regulatory changes and technological advancements.

ACT: How are recent regulatory changes allowing for more advances with technology and AI?

Jones: The short answer: they haven't. That's the short answer of it is they haven't really embraced this idea of AI. They're talking about it now. They're talking about these new technologies, but we're still in a world where we don't want to commit to this idea of with a TMF, you have to manage it electronically. You can run an entire trial today in paper and the regulations for that, the requirements for that, are down here, whereas if you say, “Oh, I'm doing it electronically,” they're up here. The simple answer is they haven't, but that doesn't mean that they've also said we can't use them. What they've started to do is tease with that idea of AI coming in, tease with that idea of new ways of working, and I think our industry is ripe for opportunity. Within our TMF service group alone, we're managing thousands, tens of thousands documents a day. You scale that up by weeks, months, years, we're talking millions of data points. That's ripe for AI, that's ripe for all these things that we've seen and I think the thing we need to look at, really, is just sensible ways of how to use it. On one side, we can really look at having a look and saying, “Okay, what about automation? What about processes and new solutions.” Electronic TMFs were an interesting new technology in that the fear mongering of the 90s was computers were going to steal all of our jobs. I actually feel there's probably more jobs in TMF today because of the electronic TMFs than there were when it was paper. We've already brought in different ways that we're using AI to help identify records as they're uploaded, and give suggestions to people of: “We think you should file it here, or this looks similar to these documents you've filed over here.” Using it in that way has those benefits for us, and again, there's nothing that says we can't utilize these.

The second biggest area is big data. Again, millions of data points here, that's before we even start coming into metadata and all the other stuff that we collect, so these things that are coming up now in in a lot of other industries, of AI agents, being able to ask, “How's this going in this country? What am I missing over here?” and having something that has that holistic view that can guide you, it's not going to replace anybody, by any stretch, but it's definitely going to help us. I think really, there's a real opportunity here for someone who's willing to push that boundary, someone who's really willing to be a first mover. AI isn't scary, the current trend going around that I've seen on LinkedIn is everyone's turning themselves into action figures. That's not scary. There's fun to be had, and there's interesting ways we can use it. I think the thing that we're going to have to bring in, which is the scary part, is a lot more logical thinking. A lot of the time we've been used to old ways of computerized system validation, which is: “I've got requirement, I have got an expectation of when I push this button. This happens.” Can't test AI in those ways. Some things you can, but there's a lot of things we can’t, so it's going to be coming up with new ways to test it, new ways to utilize it, obviously, thinking about the security and the privacy and those things are going to be paramount for us in this industry and clinical trials, but there's real opportunity here. Again, although the recent regulations don't suddenly open flood gates and don't give us a step by step on how to use it, they also don't say we shouldn't use it, and they also don't prohibit it, and I think it really is waiting for that person to take that step forwards and say, “No, this could make a difference,” and go for it.