RadiantGraph CEO: Health Care Is Just Scratching the Surface of AI

SPEAKER_1 (0:00)

Anmol, thank you so much for your time, good afternoon to you. So we heard from UNH today, United Health, obviously has been investing a lot in AI. They say they're seeing the benefits. So where are the areas that you're seeing the most spending and the most benefits across this sector?

Anmol Madan (0:16)

Yeah, and thank you for having me. I think we're still very much in the first innings of AI in this sector. A lot of the investments inside larger health organizations have been focused on operational items, member operations, prior auth, on the health system side. It's taking on the front office, scheduling, intake, clinical note taking during clinical sessions. So we're just really seeing the first step, the operational elements of health care really be adopted more broadly. I think that's just a start. I think we're just crashing the surface. There's a lot more that has to come after this for AI to truly have that impact in health care.

SPEAKER_3 (0:55)

Anmol, as you mentioned that we're in early innings here, realistically, how long could it take for these investments that are being made by health care providers to actually show up in measurable cost savings?

Anmol Madan (1:07)

Yeah, I think some of the short-term operational improvements should be showing up in cost savings in the next three months, six months, nine months a year. For these investments to make sense, they have to fit in the operational plan if not for 2026, but definitely for 2027, because when done correctly, they should be replacing direct spend that's either on business processes, or lost payments, or perhaps human workflows that are falling between the cracks that these processes address. I think the operational stuff, ideally, if you're making investments, three, six months, nine months. I think the tricky part though is, are these the right kind of investments, and are these actually production-grade deployments, or are these just investments in AI usage, which may not have that short-term return on investment if it's not pointed correctly?

SPEAKER_1 (1:58)

Anmol, two things I hear a lot about the American healthcare industry is that it's, one, confusing, and two, costly. Can AI address both of those things?

Anmol Madan (2:10)

That's a great point. So my side gig is, I'm the author of The Machines Will See You Now, published by Johns Hopkins Press. I talk a lot about in the book of the longer-term trajectory of AI. So sadly, in the short term, most of the investments in AI, and there's multiple studies, the Blue Shield Association had a study, other industry experts have written several other studies. They're actually driving up the cost of healthcare.

Most of it is finding things that perhaps providers weren't reimbursed for or perhaps reimbursed correctly and addressing those. That's kind of the short-term rise in healthcare that's perhaps influenced by the adoption of AI to a small degree. I think there's a longer conversation and a longer-term trajectory of how do we make healthcare more, the impact of AI in healthcare more deflationary, which is what we've seen in every other industry at scale with increasing automation, with increasing adoption of technology. Part of my concern personally is I think we're not spending enough time thinking about that part, and we're very focused on the short-term wins that in some cases might even increase the cost of healthcare.

SPEAKER_3 (3:21)

So Anmol, you bring up a good point there, that a lot of these are creating efficiencies and clearing more time, but there may also be leading to additional tests, which of course we know are costly. Sam, if you have had experience now with the American healthcare system are costly. How do you make the distinction of AI in healthcare about it being an efficiency play versus the more broader concern or the more broader care, in lowering the total cost of delivering better patient outcomes? I mean, how do we balance those two things?

Anmol Madan (3:56)

Yeah, and that's a great comment, Simon Roshan, and I think there's two parts to making AI deflationary actually reducing the cost of care. And the first is, of course, finding safe, clinically appropriate, validated venues for AI based agents and AI based systems to play a slightly bigger role in how care is delivered in the country today. So an example is, a lot of people who are on, say, Medicare, may only see a specialist for a chronic condition, perhaps an MSK condition, a behavioral health condition, once every few months, perhaps once every year.

So can these AI systems play the role of care managers, perhaps care guides and guide people through that journey and help them manage their condition, keep their lifestyle, their diet, their behaviors, their day-to-day activities, and manage them in a way that actually reduces the severity of their chronic condition or their mental health needs. So these are all well established in clinical literature. But one of the puzzles here is how do we make the AI systems kind of allow them for them to do that. And then the second puzzle, of course, is how do we make sure that companies are incentivized or health plans or peers are incentivized, large employers are incentivized to actually enable those and pay for those kinds of use cases, which is kind of part two of that puzzle.