Delivering Better Outcomes for Patients with LLMs

ACT: Where do you see the use of large language models (LLMs) in clinical trials in five years?

Kondru: The mission of the vision statement at Veridix AI, which I'm the CEO of, is where human intelligence meets artificial intelligence. We have a strong belief that, at least in the clinical space that I'm looking at—AI is going to augment everything that we do every step of the way, not by replacing human beings, but by actually helping them to deliver better outcomes for patients and participants across the globe. I see that there'll be a deep impact on probably seven areas that we are looking at. We talked about trial design and execution, that's one big area that will optimize trial protocols, identify optimal patient populations, and predict trial outcomes. The second area is enhanced patient engagement. We talked about connection with social media. AI-driven chat bots will provide comprehensive patient support, education, and trial participants and trial management with this participation of the patients. Third area, we talked about ethical considerations and bias mitigation, right? Researchers will focus on LLMs to make sure that there are diverse patient populations. Diversity is a must in clinical trials. Fourth area is regulatory compliance. As you know, it's a fairly regulated environment. LLMs hopefully will streamline regulatory submissions, audits, and enable quality in all of the data that we collect. Advanced data analysis instead of finding out there is an issue with the data set—we could actually find it out in near real-time about that particular issue and be able to address it immediately, thereby imparting safety for everything that we do, decreasing the number of safety events is another area which we are looking at, so decreasing adverse events, or serious adverse events, for the patients. Eventually, my hope is truly, anyone across the globe will be able to participate in any clinical trial through this decentralization.

I want to go back to the first statement I began with: clinical acceleration and decentralization are the key aspects of how we can use LLMs to accelerate the entire clinical process, be it for Phase I, Phase II, Phase III, or Phase IV, or investigator-initiated studies, but all parts of this clinical study can be accelerated. We are looking at, as I said, anywhere—text to text, text to table, text to figures, text to listings—transformations happen to ingest some artificial intelligence and LLMs to provide meaningful outcomes for patients.