IQVIA's Global Trends in R&D 2024 Report

The report mentions various strategies used to improve clinical development productivity, including novel trial designs and decentralized methodologies. What are the potential benefits and limitations of these approaches, and how do you see them impacting future clinical research practices?

We know that all of the sponsors of clinical research are looking for ways to improve their productivity. We included in our report, what we call the clinical development productivity index, it's simply looking at success rates, divided by complexity, and multiplied by trial duration. And that's a sort of simplified metric, if you like of productivity, but it's one that we know that everyone is trying to see improve, we do actually report an improvement in productivity in 2023, which is very heartening. And much of that is due to an increase in the composite success rate of molecules moving from one phase to the next that we saw come through in 2023. But the sort of approaches that sponsors are applying to improve their trial productivity, there's a growing number of them that are being used at some level that are all designed to shorten trial time to reduce the complexity of the trial. And indeed, to reduce its cost.

We see things like novel trial designs or decentralized methodologies or use of biomarkers use of prescreened patient cohorts, all as ways to try to achieve that goal of faster trials faster recruitment and enrollment, fewer steady purchases. And dropouts, less whitespace, between trial phases, better data collection and more extensive data collection, and so on. And, you know, a lot of the technologies and approaches, you know, are able to, to achieve that we also see real world evidence being incorporated into more trials, particularly, as compared to arms, or to provide natural history. baselines for comparative purposes, we've seen a lot of effort by the FDA to embrace the use of real-world evidence and to issue their guidance as to, you know, how that real world data should be gathered, interpreted, and submitted to the FDA as part of approval packages. So, again, the broad benefits are all around time cost, you know, quality of the data that gets gathered in the context of the trial. I think the limitations we see, a lot of those are really around the conservatism that exists, among many sponsors, in terms of trying new approaches, in the area of decentralized trials.

We've seen quite a bit of movement over the past few years, as sponsors have gained more experience with where those work well, and where they may not, may not work so well. So, some sort of recalibration of the way in which decentralized trials are actually conducted. But, you know, overall, there's, there's a steady movement towards some of these innovative approaches. But again, this is a, this is a conservative space, I would, I would argue, and, and some of it is difficult, it's difficult to execute. You know, it's easy to talk about using war biomarkers, but you first have to identify and validate the biomarkers. And that's not always so easy. So, some of it is simply, you know, this is not a straightforward path to being able to adopt some of these productivity enablers.

Despite increasing investments in AI for drug discovery, the report mentions that no novel active substances have yet been brought to market through this technology. What are the key hurdles hindering the translation of AI-powered research into tangible clinical benefits, and what future developments might be needed to overcome these challenges?

So, we are all excited about the potential for artificial intelligence and machine learning to transform all of our lives. And not least, in the area of drug discovery and clinical development. We are following the sort of cohort of molecules that we can see, were originated by AI research companies, about 30 trials per year that we've been tracking the last two or three years or so they are working their way through clinical development. None of them has come to market yet that we're aware of. And again, not every company discloses the sort of technology that underlies their drug discovery efforts, but best we can tell. We're still a little bit shy of an AI discovered candidate actually reaching the market, but we're moving towards it. That's clear.

I think the main opportunities we're seeing for AI lies in drug target identification, which is really about applying AI to vast omics datasets of one type or another to characterize disease states and identified novel druggable targets. We also see AI being used to design the drug. So that's where AI is being used to analyze molecule’s structure, molecular dynamics, and so on and actually designed the drug candidate itself. And then we also see AI being used to perform trial simulation and which is also a way to help optimize clinical trial design. And we're seeing that being applied to some of these new drugs that are coming into the pipeline. We also see AI being deployed as trials are being executed, in particular, to identify optimal sites, investigators and target patient populations, for potential recruitment into studies. So, we do see a lot of activity there, again, all focused on trying to achieve enrollment in shorter timelines, with more eligible subjects being screened and fewer dropouts along the course of the trial. So, you know, I think we're still in very early stages of the AI revolution.

It's a crawl, walk, run progression. And, you know, we're definitely crawling. We're, we're almost walking, we might say, but I think we're going to see more progress over the next several years before we can really declare definitively the magnitude of the positive impact that AI and machine learning is having on the clinical development and innovation ecosystem. Which means, you know, more drugs being developed faster at lower cost, leading to better treatments for more patients. Right. So, if that's the goal, we're still a little bit. You know, we're still a few years away from that goal fully being achieved.