Tom Lawry and Eric Topol
Tom Lawry and Eric Topol

TOM LAWRY: What advice would you give to a practicing physician on the front lines when it comes to how they should approach this new revolution in large language models?

ERIC TOPOL: It’s funny you bring that up Tom, because I was in clinic just yesterday with my fellow, and at the end of clinic, I sat down and said, “Andrew, you know how lucky you are? You’re going to be practicing medicine in an era where you’re not going to have to type on a keyboard, where only your conversations about the physical exam will be captured and you’re going to be making synthetic notes that previously you would have toiled over.

“But more than that, all these other things, whether it’s the preauthorization, it’s the follow-up appointments, the billing, the labs that you want to order, the next appointments, the nudges to patients and so on.” I said, “we have so much administrative work when we simply ask a patient to check their blood pressures and request that they send us the data so we can adjust their medication to find the one that works and the dose that works.

“It’s all going to be done for you. That is, checking in with the patient and getting those numbers, and then checking back with you as to what you want to do. So you’re so lucky, Andrew. You get to practice medicine.” I said, “I wish I was you. And that you’re going to have help with diagnoses, and your patients are going to have a layer to approach before you.”

Now, having said all that, and this really happened yesterday, he was smiling and he’s about 30 years old. I’m twice his age and I wish I could go back and practice medicine in the era that’s coming. But obviously, it all has to be validated. It all has to be shown that the benefit is greatly overriding any potential drawbacks.

But it will happen. It’s just a matter of time now. And so it’s exciting. It’s just so damn exciting really.

TOM LAWRY: I think about what economists talk about, which is the exponential growth curve, and I think the world, particularly health care, is used to managing linear stable change. And I think we may be approaching this exponential growth curve. And as that happens, the capabilities are great but the ability to assimilate it, bring it into the practice of medicine in these big health care organizations, I think that’s going to be one of the biggest challenges faced by health care leaders today.

ERIC TOPOL: No, I agree. I think, in general, medical change is very hard to come by. It’s a sclerotic, almost ossified type of situation as we’ve seen. We have clinicians who are overwhelmed and hate to be data clerks, and they need to be rescued. And on the other hand, we have patients who can’t reach their doctors and can’t get appointments.

They need answers, and they don’t really want to go to urgent care and emergency rooms if they can avoid it, of course. So I think we have a remedy across the board. Being able to deal with a patient’s data specific to that person, not like a Google search as it exists today. So this is why we’re looking at something we’ve never had before.

I do think the medical community change is more likely to occur now at a faster pace than previously because it’s essentially providing some relief that’s just been building up over a long period of time. And in order to provide the best care for patients, what we need to do is take the things that can be automated accurately.

Let’s give patients more autonomy when we know it’s reliable autonomy. And with that, we get to a reset of how medicine is practiced, and we get to a happier state on both sides, clinicians and patients. And so the faster we get there, the better. The problem we have though, Tom, is the medical community isn’t well versed in AI in terms of its nuances and capabilities and limitations. There’s no curriculum in a medical school today that teaches AI.

There is no change in who we should be selecting as the physicians of the future, but we don’t need to be brainiacs, but rather need to be really great communicators and people with empathic features and interpersonal skills. So it’ll take time to get there, but I’m confident we eventually will.

Also of Interest