Accelerating Clinical Trial Timelines with AI

In a recent video interview with Applied Clinical Trials, Gaurav Agrawal, Senior Partner at McKinsey & Company, highlighted insights from the company’s new report on clinical operations. Agrawal outlined how AI can accelerate development timelines by optimizing trial design, reducing operational “white space” between phases, and streamlining study startup, monitoring, and closeout activities. He emphasized the value of human-AI collaboration, noting that AI provides insights for protocol design, site selection, and monitoring while humans contribute strategic oversight and relationship management. Agrawal also highlighted the growing need to integrate trials into routine care and expand access through trial-naive sites and simplified protocols, as well as the role of digital tools in decentralizing assessments and enhancing patient-centricity. Looking ahead, he stressed that preparing for end-to-end AI-enabled trials will require systematic adoption of technology, cultural and workflow transformation, and cultivating digitally savvy clinical operations talent.

The below interview transcript was lightly edited for clarity.

ACT: The McKinsey report suggests AI could halve development time. Where do you see the biggest near-term impact on clinical trial timelines?

Agrawal: Yeah, that's a great question, Andy, you know, I'll answer it in two chunks. One is sort of, what, where do we see AI impacting overall, and where do we see it more in the near term? I think there's actually a lot of areas across what I call the development value chain, where AI could actually have an impact. And the three or four that are top of mind for us are upfront, as you look at AI using big data and the different data sets in actually designing better clinical studies, studies that will not require as many patients to generate the same amount of evidence because you powered the studies more through AI, as well as studies that will recruit faster.

So there is one element of getting to better trial designs that can recruit faster and maybe have fewer patients than what is needed in a conventional randomized clinical trial. The second area, we do think there's a potential to reduce the white space, so the space between, let's say, a finish of phase two to a start of phase three, typically anywhere between six to 12 months. And it includes everything from decision making on phase two, the new protocols, and sort of getting ready, gearing up for the next phase. And I do think there's a lot of potential there, especially in terms of design of protocols, in providing insights that give us better conviction to take some of the decisions where a lot of time is lost between debate.

And so I think there is an element of cutting out some of the white space between the timing, if you will. I think the third thing that is related is sort of the more operational aspects of study startup. So a lot of manual work, document generation, and build of databases, EDC, etc., goes on, which, in a conventional and traditional way, happens mostly through humans and manual effort. And I think a lot of these activities could be replaced or augmented with AI, if you will, right?

And then the fourth one, as we get into the study, even sort of thinking about, what sites do you actually go to? I think AI can power a lot of insights around selection of better clinical trial sites that will recruit faster for my protocol and for my particular drug as well, as you know, guide the monitors, where should the monitors be actually spending more effort versus less in terms of their monitoring activities, and then, of course, at the tail end, sort of the activities of database cleaning and document submission, et cetera.

So if you look at the overall value chain, as I call it, there is a good potential to shave out sort of six to 12 months from a design and sort of better designs, et cetera, and meaningful portions also through the operational aspects. And when you add this all up, it actually is a pretty significant impact.