October 10, 2024, 10:04 AM EDT
(Originally aired October 1st on Clinical Research News Online)
________________________________________________________________________________________________________
Read More on How Causal AI Helps Diversity in Trial Planning
In this episode of The Scope of Things podcast, host Deborah Borfitz engages with Aaron Mackey, VP of AI and Data Science of Lokavant, who talks about the unintended consequences of decisions made during trial planning that can lead to questionable conclusions, how AI and ML are helping with the diversity issue in trial participation, and his stop-gap emergency plan to keep trials on track if there is no digital support available.
Unintended Consequences of Decisions Made During Trial Planning
Aaron Mackey begins by addressing the overarching issue of confusion over when study variables are causative versus when they are simply correlative. He underscores the difficulty of distinguishing the differences between them and gives an example of an instance in patient recruiting during trial planning that leads to questionable conclusions. These conclusions lead us to think about where we might be led to make a change, a protocol amendment, even to that eligibility criteria. We have to ask ourselves, what is actually going to improve the metric in this specific case?
How AI and ML are Helping the Diversity Issue
Deborah Forbitz then questions Aaron in the next topic of how Lokavant is using AI and Machine Learning to improve clinical trial diversity, as is now mandated by the FDA. Mackey starts by noting the mutual understanding among researchers that conventional methods of patient enrollment formats is not a very reliable method, and adding Causal AI with a human element to the mix helps to eliminate bias in site selection and have more accurate projections.
Mackey also underscores the point that it's not just about the enrollment rates, but the overall protocol design itself that needs AI and ML optimization. What are the eligibility criteria, the patient burden, and the site complexity, and how do all those things bias? Who can actually participate in the trial or not? AI and ML optimization in trial planning is crucial to addressing these concerns.
Aaron's Stop-Gap Emergency Plan
The conversation lightheartedly pivots to a question that must be on the mind of all tech-weary types: what would be your stop-gap emergency plan for keeping clinical trials on track with no digital data or AI at your disposal? Aaron Mackey hones in on how digital data and AI efficiently transmit data in clinical trials. However, he reminds Deborah and the listener audience that the part the "deserted island" question misses is the human-to-human, interactive element that is imperative and can be overlooked when focused too much on the digital connectivity of all the various participants involved. Mackey believes that the best AI is still human in the loop.