Current Uses of LLMs in Clinical Trials

ACT: What is the current state of the use of large language models (LLMs) in clinical trials?

Kondru: I think there is a lot of interest in LLMs in clinical trials. However, we see adoption has been low due to privacy and security concerns. It's a regulated environment, as we all know, and the second one is the specialized expertise to implement AI (artificial intelligence) solutions within the clinical landscape has also been limited in many organizations. That's one of the barriers for a lot of organizations to adopt LLMs within the clinical space. As you know, the adoption has been slow, but what we have been doing is to take decades of experience in using neural networks and semantics and so on—some of the specialty technologies—to implement a platform-wide approach, what we call the Veridix AI platform for clinical trials. Hopefully this will increase clinical acceleration, also clinical decentralization, which is giving access to more patients across the globe.

I see there is also a concern on ethical considerations and thinking about what is the impact of AI on participants, patients, and so we have to be very cautious in terms of using some of these technologies. You have to fully leverage them, use them, but with the prudence that is necessary in every clinical trial, because it affects human health. The areas that I feel like really important—I talked about clinical acceleration and decentralization—things such as patient trial matching, clinical trial planning, and execution. We are seeing a lot of use cases that pop up in improved efficiency for all of us who are in life sciences to execute clinical trials across the globe, enhanced quality of the information that we put out. I also see there is an area that's growing quite rapidly, which is the therapeutic area specific LLM use. So be it in neuroscience, be it in ophthalmology, be it in vaccines, the needs are different for participants, and LLMs are coming to that risk, if you will, in addressing some of these compelling needs for clinical trials. So certainly, the two concepts to remember is you'll have a profound impact if adopted well in clinical acceleration and decentralization.