Artificial intelligence has rapidly become a buzzword that evokes fear and hope. A recent panel discussion dug into this issue at the 2023 BIO International Convention, one of the world’s largest biotech partnering events, attracting more than 14,000 biotech leaders to Boston this year.
The discussion, “Is Your Billion-Dollar Oncology AI Cutting-Edge Technology or Is It Artificial Ignorance?” explored how artificial intelligence (AI) could revolutionize cancer drug discovery—or become a tool for amplifying biased data.
Karim Budhwani, Ph.D., DLA, CEO-scientist of CerFlux, Inc., moderated the panel, which consisted of Maggie Scully, Ph.D., Partnership Development Office director at the Frederick National Laboratory (FNL); Sandeep Menon, Ph.D., chief scientific officer of AI and Digital Science at Pfizer; and Tahera Kan, M.B.A., M.S., vice president of Precision Medicine and Enabling Technology at Johnson & Johnson.
Promise of AI
AI–which refers to computer programs that use data instead of programmed instructions to make decisions or predictions--is currently being applied in cancer research and has shown tremendous potential to accelerate cancer drug development, advance early detection of cancer and improve personalized cancer care.
However, Menon explained, “It is not a silver bullet that can replace the expertise and insights of our scientists. While embracing this technology, we cannot lose sight of the human element that drives scientific discovery. By combining the strengths of AI with human expertise, we may be able to unlock unprecedented advancements and benefit patients worldwide.”
The group also reiterated the need for AI, among other tools, to accelerate the slow pace of drug discovery.
“Drug discovery can be a 20-year, billion-dollar process, and it can still fail at the end,” Scully said. “What does that mean for patients?”
Data and collaboration
The panelists emphasized the need for high-quality data for training useful models.
“Racial and ethnic biases have come to surface as a critical shortcoming of analytics and artificial intelligence in virtually every sector of endeavor, due to skewing of data due to disproportionate disparities in these datasets,” Budhwani noted.
However, Kan explained that this does not have to be the case in oncology. She said, “Partnering with others and using a consortium approach can help increase the chance of success.” She advocated for “joining efforts in a pre-competitive space to ensure the diversity, quality, relevance and sufficient size of the training data set” to develop models that deliver meaningful insights.
“There are opportunities for consortia-building and partnering to support success,” Scully added. “At the FNL, we aim to democratize resources to help tackle these problems.”
One message was clear from the discussion: there must be balance in how AI is used and regulated. Regulation, while critical, should not hamper innovation.
Panelists described a need to inform regulatory frameworks and noted that the Food and Drug Administration released a discussion paper and request for feedback on AI and machine learning in drug development. They said it is now up to the community to give input.
While data are important, Scully noted that the predictive values are critical for evaluating AI tools. To aid with this, she alerted attendees to resources—cutting-edge computational models, data sets and algorithms—that are available to the research community through a National Cancer Institute–Department of Energy project, which is facilitated by the FNL.
Ahead of the panel, Budhwani spoke of this need for open dialogue as a reason for organizing a discussion on this topic.
“We have an incredible opportunity here to shape the future,” he said. “Because we have so much potential here, we have to lead it. That’s why it’s important to engage now, engage openly, and engage with a lot of people—because this is how we can shape the future.”
The crowded room clearly demonstrated that conference attendees agreed with the importance of this discussion.
“It was incredibly energizing to see the vibrant innovation ecosystem as we walked the halls at BIO and think that AI and machine learning will enable the next wave of innovations for a brighter future for patients,” Kan said.