SCOPE 2024: Wendy Morahan of IQVIA Discusses the Use of AI in Clinical Trials

Applied Clinical Trials: How do you think artificial intelligence (AI) can advance clinical trial execution?

Wendy Morahan: It's a big question because I think there are so many ways, from preclinical drug discovery all the way through post marketing of pharmaceutical treatment of the drug. In clinical trials, I think it's everywhere from planning the study, doing things like protocol writing, protocol strategy, doing things like the myriad of plans that are created when you're starting up a study, the statistical plan, the data management plan, the monitoring plan; these are all kinds of things where you can use generative AI to create the content very, very quickly, and then have the human adjust that content, make sure it's appropriate for the study, that it's accurate, and that it's high quality. I think during the conduct of a trial, there are ways for looking at generative AI to check data discrepancy that's coming in from the various places where we capture data, whether it's EDC or connected devices, the ability of the machine to look at the data very quickly and very high volumes and maybe point out trends and discrepancies to data managers and data reviewers and medical reviewers. I think one thing that's really gaining a lot of attention this year, of course, is generative AI with all the ChatGPT stuff. So using generative AI to be able to have a conversation with your data, just ask questions with your own voice of the data and get almost an immediate response. So interrogating the data in a much more freestyle, human like way. All of that is going to add to speed, efficiency, quality data transparency. I know that's a very long answer. But there's just so many ways that AI is going to impact clinical trials.

ACT: As more tech providers move into the clinical trial space offering solutions with AI, what do decision makers need to be thinking about when choosing solutions from these providers?

Morahan: There are a lot of things to consider. And I think that when you're talking to a technology vendor, you want to look for people that are thinking about what it means to use AI. So how do you responsibly deploy AI into your systems? What are you doing around data privacy? We all love talking about ChatGPT and Copilot, Microsoft has Copilot and so many of their tools now. Well, that's like a public domain large language model. That's not actually what we're doing when we use generative AI in the clinical trials setting because we can't ask a question that's going to be then saved in a public domain large language model. We have segregated, more private, more protected, behind firewall, large language models that need to be used when we're using generative AI. So are they doing that? Are they focused on the privacy of people's information? Are they focused on the ethical and unbiased use of AI? So you hear people talk about that you can actually exacerbate existing bias by using datasets that have bias in them. So are you looking at making sure you're using unbiased data in order to have unbiased answers? And are you protecting against things like hallucinations? And you're basically getting the wrong answer from the model. And are you developing AI that is explainable? And are you surfacing that? Not just saying that it's explainable, and you've written down what's happening, but you can surface it, and you actually know what the model is doing at any given time. So I think looking for companies that are making commitment to how AI is used, and they're willing to talk about it. I think that's the first thing to look at.

ACT: What potential application of AI in trials excites you the most as we look forward in 2024?

Morahan: A little bit biased here for my own personal passion, I'm in clinical data and analytics and I'm squaring that middle of the conduct of a clinical trial. So I'm really excited about the conversational AI, the generative AI during the conduct of a trial, the ability to ask questions of your data on the fly, ask questions, like I said before, in your way; in your language, interrogate the data, get an answer back, maybe ask the next question, and truly have this dynamic conversation with your data; to learn about your data, to learn about what's going on in the clinical trial. And I think that's going to be something that just becomes kind of expected in these technology tools that are holding all of this data, so I'm really looking forward to doing that this year.