Using AI to Individualize Clinical Pathways Is the Future of Precision Oncology: Joshua K. Sabari, MD

Joshua K. Sabari, MD, director of High Reliability Organization Initiatives at NYU Langone Health's Perlmutter Cancer Center, discusses how clinical pathways can be helpful for doctors looking to give their patients the best care as well as how those clinical pathways may be improved with the use of artificial intelligence (AI).

This transcript has been lightly edited for clarity; captions are auto-generated.

Transcript

What role do clinical pathways play in your day-to-day when personalizing oncology care?

Pathways have become critical in thinking about how to guide care for the individual patient. Imagine if you're treating colon cancer and breast cancer and lung cancer and pancreatic cancer: it may be hard to stay up to speed on what is the best frontline therapy, or what is the best frontline therapy in a patient with a specific alteration, let's say KRAS G12D. Pathways really help us streamline care. They help universalize care across large health care systems.

Pathways sometimes are cumbersome. They're generally built into our electronic medical record. They require us to click through many prompts explaining the rationale for the therapy that we're giving that patient. And it's important to note that pathways are meant as a guideline, as a sort of bumper in that bowling alley. We can sometimes go past them and go outside the lines, and it's not wrong, essentially, if you have an option for a patient that is outside of the guidelines. But again, in clinical practice we want to make sure, especially across large systems, that we're really focusing on giving universal care to patients. You may have a site in Long Island, you may have a site in Brooklyn, you may have a site in Manhattan. You want to make sure that all patients are receiving the right care across the continuum.

What do you think is the future of precision oncology?

I think we always try to individualize care for patients in the interactions that we have, and getting to know the patients on a personal level, getting to know their families. What makes the patient tick? What are the patient's goals, and what do they want to achieve from their cancer care and their cancer therapy? I think on a deeper level, we talked about the biology of the cancer itself and how no 2 patients are the same. Patients have different genetic alterations, different protein expression in the tumor. I think what we talked about is, into the future, using AI platforms to help us better guide systemic therapeutic options and also to help us better sequence therapies. Many patients with different cancers have multiple options for therapy, and we don't know if option A is necessarily better than option B or option C. I think, into the future, if we can start to collate many cases—thousands, hundreds of thousands of cases—we may be able to better match and guide therapy. Maybe in the future, pathways might not only be what is the best therapy here, but what is the best sequence of therapies for this individual patient based on their individual history and sort of genetic makeup?