Image Credit: Andy Studna
A panel at the Summit for Clinical Ops (SCOPE) 2024 in Orlando, Florida discussed where industry currently is with utilizing generative artificial intelligence (AI) in clinical trials. The panel was moderated by Brian Martin, research fellow & head of AI, AbbVie, and the participants were Neil Garrett, head of regulatory medical writing, Johnson & Johnson, Hoifung Poon, PhD, general manager, health futures, Microsoft Research, Prasanna Rao, head, AI & data science, data monitoring and management, clinical sciences and operations, global product development, Pfizer, and Samar Noor, vice president, head of statistical programming, global biometric sciences, Bristol Myers Squibb.
Martin kicked off the conversation by asking, “What can we do?” Poon began by answering with some things he is excited about in the field of AI, specifically looking to the future and how AI has the potential to predict disease outcomes. Poon also emphasized that AI should serve as an assistant to humans, rather than completely taking over important tasks. Rao then followed by describing how far AI has come in recent years to the point of automating study protocols and generating use cases, which could not have been done several years ago.
Garrett discussed the use of AI with human intervention. He mentioned that there is a solution for many elements of a trial now such as patient recruitment and evaluating patient burden. He then emphasized that industry should not be letting AI alone make decisions on managing studies, but rather finding a middle-ground where AI is utilized, but humans make final decisions. “That’s the area we see the greatest opportunity,” Garrett said.
Noor also looked to the future and explained that improving the predictability of industry’s current models will only help improve AI processes moving forward. There is also great potential for AI in precision medicine, which will greatly benefit patients. However, Noor also emphasized that it is important to remain cautious in this space because datapoints in any given study can often be biased, leading to inaccuracies when using AI.
Overall, the panel shared the main sentiments of being excited for the continued use of AI, but also remaining cautious and emphasizing that AI should never completely remove human intervention.