Overcoming Barriers to AI Adoption in Precision Oncology: Davey Daniel, MD

In an interview, Davey Daniel, MD, chief medical officer of OneOncology, discusses how artificial intelligence (AI) is being applied in precision oncology workflows. In contrast, he also addresses barriers to its integration and offers potential solutions.

Daniel expanded on these themes earlier today at the Patient-Centered Oncology Care conference in Nashville, Tennessee, during the panel discussion, "Access for Everyone: Using AI in Precision Oncology," alongside Alyssa Schatz, DrPH, MSW; John L. Villano, MD, PhD; and Stephen Speicher, MD, MS.

This transcript has been lightly edited; captions were auto-generated.

Transcript

In your experience, how is AI currently being used to support precision oncology workflows?

We're really in the early days of bringing AI into oncology care. Right now, there are a lot of applications that are very promising. There are a few on the operational side that are already embedded in clinics: using AI to better understand business and to execute more efficiently.

Clinically, I think physicians are really embracing AI as an extension of their knowledge, particularly OpenEvidence and other features that help us really stay up to date. I think many of us are embracing the use of AI in taking notes and, really, in the clinic flow.

I think the next step is to really figure out how these tools are actionable for patient care. That means really operationalizing a huge amount of clinical and molecular information to help identify therapeutic options and identify trial options, which I think is one of the biggest areas. Yet, we've got to figure out what those guardrails are [that are] needed to protect patients and really support the physician judgment.

What are the biggest barriers to integrating AI into precision oncology? How can they be addressed?

I think one of the biggest barriers for AI in general, but also in precision oncology, is building trust. Physicians, patients, and practices have to have confidence that the information produced is really reliable. That probably means making sure there's transparency on how the models are built, which data is relied upon, and how they reach their conclusions. That means regulation is important; we have to kind of cross that threshold.

I also think one of the biggest challenges, particularly for our community practices, is ensuring that practices learn how to evaluate these technologies and how to evaluate the partners they're working with so they make good choices. [There are] a lot of people out there in the field, so we have to make sure that we're partnering with the right groups.