Integrating AI Into Clinical Trials

ACT: Where does the clinical research industry currently stand when it comes to integrating artificial intelligence (AI)?

Stephen Pyke: Where are we seeing the uses [of AI]? it's certainly been used a lot in discovery, understanding human biology, it’s been used a lot in chemistry, and helping choose the right molecules. It's being used in the clinical setting for diagnostics. And we've been hearing about incidences of using AI to interrogate X-ray, CT scans, and images of all sorts, and do so very successfully, very rapidly, and very reliably. We're all experiencing AI in a limited way in wearables. And we see wearables almost ubiquitously now, in clinical trials, I would say. And I suppose the other big area where we've seen AI make strong inroads already is in safety and pharmacovigilance, where you're using the ability of AI to read natural English language, and indeed other languages, of course. But natural language, make sense of what it's reading, and then use that to identify important new safety signals. So all of that is already happening; we're going to see all of that continue to advance. And we're also going to see a risk-based encroachment into clinical trials. But I think I've said already, I think that approach is going to be more careful and more thoughtful, perhaps slower than you might expect. I would call out a few key areas where we were going to start to see real evidence of particularly large language models and next-gen AI making real inroads. One is around, and this probably sounds a bit kind of dull, in a sense, but it's important. And that's workflow automations, operational productivity, the important stuff that that gives us the important benefits that allows us to move through the whole process of designing, planning, designing, conducting, and completing a clinical study more rapidly. So we're going to see a speed up in clinical trials, and that will become increasingly noticeable to all participants. I think over the next couple of years we're also going to see a lot of clinical trial participants using AI to support decisions; decision making. So AI, which is based on the ingestion and synthesis of large volumes of evidence, one only has to look at the number of medical journals, the amount of literature that's out there, around new drugs coming through, and it's almost impossible for humans to stay on top of that. AI does that incredibly fast, incredibly well, and allows us to make really well-informed decisions as we're planning our programs as we're designing our studies. And then the other big area is search and retrieve. And this idea that we can access knowledge and information quickly, again, allows us to ensure that the people involved in clinical trial conduct have access to the information that they need, when they need it, to make high quality decisions, the right time first, every time. I suppose what I've just pulled out, and this is getting to the end of my rather long answer. I think what I've called out is AI really enabling more clinical trials, better clinical trials, more drugs being delivered, being approved, more rapidly supported, additionally, by the already existing investments in the upstream end of drug discovery, meaning more drugs coming through the pipeline, because AI has done a better job of finding opportunities. So in that sort of context, I think AI, you could really characterize it as driving speed, driving quality for right time first more often, driving productivity, both in terms of the number of drugs coming through the pipeline, and the speed again of those studies when they reach us. And in aggregate, I think could be quite transformational. But when you ask the final quick piece of this which is: What will it mean to the patients?