Assessing the Promise of AI in Oncology: Looking to the Future

,

In this final episode of OncChats: Assessing the Promise of AI in Oncology, Toufic A. Kachaamy, MD, and Douglas Flora, MD, LSSBB, FACCC, discuss a roadmap of artificial intelligence (AI) advances in the next 5 to 10 years.

In this final episode of OncChats: Assessing the Promise of AI in Oncology, Toufic A. Kachaamy, MD, and Douglas Flora, MD, LSSBB, FACCC, discuss a roadmap of artificial intelligence (AI) advances in the next 5 to 10 years.

Kachaamy: So, you’ve given us a roadmap. Can you tell the listeners about your vision of where you see us a year from now, 5 years from now, and 10 years from now, [as it relates to] the practice of precision oncology and AI?

Flora: Okay, I’m gonna put on my futurist cap, because I think [we’re] moving so fast that even the builders don't know. Even the people at Open AI that built chatbot GPT 4 were not sure because these systems are iterating on their own at levels that we can’t do. I can make my best-educated guesses. Year one. This year, tomorrow. It will be an incorporation of AI tools into everyday practice and probably the first salvo will be things that improve efficiency, accuracy, and pattern recognition, so AI and radiology, AI and pathology to help us in diagnoses, AI-driven predictive analytics to help us with treatment decisions, and then all of the mundane stuff, the administrative tasks trying to reduce burnout and get the doctors back in the room.

In the next 5 years, I think we move into a new realm, where we really start to see the promise and potential in precision oncology being more fully realized. This is looking at these large omics that I just referred to in my prior answer, where we can start to really enable personalized cancer treatment with an n of 1, these basket trials, looking at a person who fits this epigenetic and genomic and phenomic appearance can be rolled into the correct clinical trial in real-time to give them a drug that might hit their FGFR2 receptor and augment their gut biome and augment their ability to respond to some immunotherapy, all in combination. That’s probably in the next 5 years.

Then, what we’ve talked about before, the 10-year goal is these major breakthroughs in our understanding of cancer. As this technology starts to self-train and start to learn on its own, it may actually unravel these genetic and molecular interactions that underlie cancer. We’ve seen this before, right? We used to think it was black bile, and then we went through a phase where we thought it was all infectious, and then we thought it was evil humors, and we’re moving to this molecular medicine era now, that is a transformational time for an oncologist.

The next step is to really, really understand where tumors come from, where they’re born, and where it is that these mistakes and mishaps happen in our DNA or RNA so that we go in and potentially prevent a repair.

Kachaamy: Wow, that’s an amazing roadmap. I’m going to take the first-year prediction and ask you some specific questions about it. For someone who wants to incorporate AI tools in their practice for next year, how do you choose software with so many startups and so many tools available out there? How are you choosing one software vs another?

Flora: You read my journal, right? We want evidence-based care. I think we need to, as system leaders, and this is the chief information officers, chief medical information officers, those early adopters, that 2.5% of people that are reading about this and who are playing with these things right now, I think you need governance. As such, I would say for systems that are starting to dip their toes into this; [you] should [have] an AI steering committee and that should be composed of clinicians, people who use business intelligence, and IT experts in your healthcare system. Or you should outsource that.

In our system, we’re starting to get those teams together again to try and figure these things out in technology committees, because you can’t buy them all and they have to be prioritized. You look at your Pareto chart and say, “What tool could I use or what lever could I pull that would save the greatest number of patients per day?” That’s my metric. We’re losing 1650 [patients with] cancer per day and my life’s mission is to find tools that accelerate discovery so that we can save half of them, or one-quarter of them, or all of them, ultimately. That’s going to take administrators, bean counters, physicians, and physician representatives to guide that discussion, and I hope we can increase oversight in legislation to make sure we do it safely and responsibly, as well.

Kachaamy: This is an area that your journal will be covering.

Flora: Absolutely.

Kachaamy: Guidance on tools. Perfect. We want to thank you for that, and I really want to thank you for taking the time to share your expertise with us today. Any final thoughts that you want to share with us?

Flora: I would say that this is an era of transformation and intense promise and optimism. I see that this stuff will be evolving rapidly and it’s going to be important for all of us to read, just like you would read about new antibiotics or new technologies in your field. This is important; this is going to save lives at a pace that we have not seen before. I don’t think it’s pushing us toward despair; I think it’s pushing us toward hope. I think I’m looking to build a future where healthcare professionals just like you and I have the time and resources we need to really provide personalized care. At that point for me, as a cancer doctor, that gives my patients the best chance of beating cancer.

Kachaamy: Perfect thank you very much.

Flora: Thanks for having me.