By Irene Yeh
November 8, 2023 | Though increasing in popularity, the application of AI is not as widespread as some may think. In the case of clinical trials, AI is seldom used to help with organizing data, managing studies, and processing information. This is a challenge that Michael Ibara, Pharm.D., Chief Data Officer at Elligo, faces in his line of work. In fact, he has firsthand experience with the reasons why clinical trials seem so hesitant to use AI.
“I’ve always been interested [in applying] technology to solve problems,” says Ibara. His curiosity led him to develop the ASTER Project, where he and his team took adverse events directly from the EHR, processed and coded them automatically, and sent them to the FDA. While this was a resounding technological success, there was a “miserable sort of failure in adoption.”
“I couldn’t get the industry to use it,” recalls Ibara. “I couldn’t get the regulators to okay it at the time.”
What exactly is preventing clinical trials and researchers from applying AI in their work? In the newest episode of the Scope of Things, host Deborah Borfitz discusses with Ibara the roadblocks obstructing AI’s potential in clinical trials and what can be done to eliminate them.
The Biggest Challenges
Throughout his career, Ibara has encountered three challenges that prevent clinical trials from further applying AI technology. The first is a lack of imagination. It sounds harsh, concedes Ibara, but there is perhaps too much focus on execution and risk mitigation, which can prevent others from seeing the full picture of the potential and possibilities AI brings, he believes.
The second challenge is an on-the-letter focus on regulations or laws rather than adhering to the spirit of the regulations. Literal interpretations of what is and is not allowed can prevent researchers and industries from achieving goals when in fact there are ways to meet these goals while still complying with the regulations already set.
And the third challenge, which Ibara believes is the biggest, is having the wrong mental model or concepts in mind. “When we did ASTER, one of the things that held us back the most was regulators and industry. Everybody pictured a piece of paper in their head, and it was just sort of very lightweight instead of data. And that prevented you from understanding that you can put in thousands of elements of data into a file that’s much smaller than a piece of paper file [or] a PDF.”
When heavily regulated clinical trials are not allowed to expand their reach and explore different paths, their ability to progress becomes limited. This not only affects the clinical trials industry, but it also affects the tech companies that are trying to enter the space, as they do not understand the nuances of regulations and processes, as well as the time it takes to go through them. This could cause tech companies to back out and decide not to partake in AI application.
Seeing the Bigger Picture and Thinking Outside the Box
“We do need regulations. The problem is…if the regulators have the wrong sort of mental concepts, they will tend to regulate one problem but actually make innovation very difficult,” comments Ibara. In other words, he wants regulators to view AI with a broader perspective instead of focusing on just a few issues or implementations.
“AI is the new electricity,” says Ibara, quoting Andrew Ng, Co-Chairman and Co-Founder of Coursera and Adjunct Professor at Stanford University. And it certainly is in terms of its broad application capabilities and potential. One example of implementation is “scut work.” Ibara defines scut work, which originated from hospitals and healthcare, as menial and repetitive tasks that are time-consuming when conducted by humans. For instance, generative AI can help with reviewing medical charts. About 20 people are needed to review 10 charts over a certain period, but AI can lower the workload to only about two people and cut time drastically. Generative AI can also help make document creation, protocol creation, signing regulatory documents, and reviewing regulatory documents much quicker and more efficient.
However, these applications cannot happen if there are no changes made. When electricity replaced steam power, there weren’t immediate productivity gains because the way the factories were run stayed the same. The factory floors needed to change first to accommodate this new power source. Similarly, in the clinical trial process, changes must be made to make way for AI.
But of course, getting people to listen and consider thinking outside the box is much easier said than done. According to Ibara, financial cases need to be made to convince people to participate, as well as to help regulators understand that the application of AI can help spur progress without contravening regulations.
Innovation and curiosity are the keys to advancing AI in the clinical studies field. To Ibara, the solution to furthering AI’s application is for regulators and researchers to be open-minded, step out of their comfort zones, and to take risks. “The goal here, I think, is to have that imagination to be able to put together what would clinical trials look like… if we took constraints off and we were able to use the technology the way it’s best used.”