As clinical trials increase in number and complexity, so does the competition for the patients best aligned to the study requirements. And with clinical trial processes relatively unchanged over the past decades, a patient and investigator bottleneck has emerged. Finally, with the convergence of artificial intelligence and algorithms trained on real-world data (RWD) sources, we are building the solutions that address this pain point.
Suzanne Caruso, General Manager for Clinical and Strategic Intelligence at Norstella, shared her latest views in a panel discussion at a recent customer event. From opportunities and barriers to redesigning studies from scratch, Suzanne anticipates the benefit of AI and RWD on clinical trials to be profound and imminent.
Q: Suzanne, what’s your reaction to the latest data on clinical trial enrollment and execution? How does RWD fit into this picture?
Enrollment rates keep declining, and I’m not surprised. Studies are more complex, patient populations are smaller, and coordinators have to juggle multiple systems just to find eligible patients. There has never been a harder time to be a study coordinator than today. RWD is a powerful tool to help, but it needs to be positioned as a way to make jobs easier and not just another layer of complexity.
Q: What do you see as the biggest barriers to using RWD and AI in clinical trials right now?
Education is key, for both pharma teams and clinical sites. The biggest pain point in a modern clinical trial is matching patients to I/E criteria via electronic medical records, coming from multiple sources both inside and outside the investigator’s network. There’s also a referral challenge: knowing which physicians are likely to refer patients to trials could unlock better enrollment, but that data isn’t fully used yet. Less than 5% of HCPs participate in trials, even though everyone benefits from clinical research. RWD and AI can solve for both of these challenges, but it comes back to education and providing a solution that reduces site burden while increasing the opportunity for participant enrollment .
Q: How is AI changing the game for clinical trial operations?
AI lets us move beyond feasibility questionnaires and investigator estimates. Now, we know how many patients come through the door and how many participants would be eligible for enrolment . That means recommendations patient enrollment can be informed by real data. AI also helps optimize investigator selection. If we can get below 10% non-enrolling sites, that’s a huge win for speed and efficiency. We can also accurately predict the effect of protocol design decisions on study outcomes via AI, which ultimately benefits operations teams. In a single sentence, AI helps us analyze patterns and predict which sites have the opportunity to perform best, while RWD provides the evidence needed to support those decisions.
Q: How do you see the role of technology evolving in clinical trials?
Clinical trials have lagged in tech adoption, but that’s starting to change. The mix of people involved is shifting: data science and tech are now part of the conversation alongside clinical trial managers, feasibility teams, and medical affairs groups. Protocol design requires collaboration across disciplines, and as we move toward more predictive and recommendation-driven development, the team at the table will look very different. I’m optimistic that this will have a long-term impact on process and outcomes.
Q: What’s your prediction for the next big change in clinical trial design?
My dream is to see ‘classic’ control arms removed from Phase III trials. If you want to change the cost and speed of bringing drugs to market, get rid of control arms. We now have the data and technological capabilities to create synthetic control arms rather than having hundreds, sometimes thousands of participants, take a placebo. The challenge is regulatory: there haven’t been many approvals using synthetic control arms yet, but that’s the direction we’re moving. Using RWD and AI to create synthetic controls could halve the number of patients needed per trial. This dramatically improves the economics of clinical development and would make many more trials possible and practical. I expect this to become more commonplace in the next two or three years.
Q: What practical steps do you think the industry needs to take to make this future a reality?
Engage health authorities early and often. Regulators want pharma to come talk to them, and there’s interest from biopharma in pushing this forward, but it’s a cost and time analysis. The data is there, but conversations need to happen to get buy-in both from regulatory bodies and manufactures, here and abroad. Companies are already building potential clinical development scenarios that involve synthetic control arms, and I think we’ll see rapid progress as regulators get on board.
Q: If you could redesign Phase III trials from scratch, what would you change?
I’d really focus on rationalizing protocols to get them right first time around. We collect so much information now that we should be able to streamline endpoints and criteria, reducing unnecessary amendments and focusing on what really matters. And, of course, I’d remove control arms wherever possible. Ultimately, it’s about putting the patient first and making participation easier and more meaningful.
