Show notes
Scot Morris, OD, and Rehan Ahmed, MD, on recent AI news, predictive AI and the state of healthcare payments.
The hosts discuss the FDA 510(k) clearance of I-Predict DR, AI software that uses the iCare DRSplus camera for autonomous diabetic retinopathy screening.
They unpack three pillars of predictive AI (awareness, desire and physical capability) and how AI could move healthcare from reactive to proactive, if the industry gets past major data challenges and human resistance to change.
Finally, they take on healthcare reimbursement and debate how AI and new entrants could modernize payment processing, cut clinical friction and bring real-time financial transparency to providers and patients.
Transcript
Welcome to Real Talk on AI in Eye Care with your host, Dr. Scot Morris. And I'm co-host, Dr. Rehan Ahmed. In each episode, we discuss current news, add in a little AI education, and debate innovative AI tools and topics that will change the industry. So this week's News of the Week, Scott, it takes us to a familiar sort of battleground or topic that we often talk about, and that's autonomous diabetic retinopathy screening. So just this month, July 8th, iHealthscreen, which is a company based out of New York, they received FDA 510-K clearances. Actually, I've not heard of this company before. For a system they're calling iPredictDR, it's an AI-powered software that detects more than mild diabetic retinopathy.
The things that are interesting here, because there are so many companies now, this camera is a camera that has not yet been used in an approval. So it's the iCare, DRS plus camera. It's actually one that I use in my clinic. No financial interest, but it's quite a nice camera. It's a tabletop that I think is pretty good. And so I guess one of the lessons we're learning is that I can't keep count anymore, Scott, with all of these new cameras. I don't even know how many there are. I try to keep a list. And I can't keep up with how many of the autonomous DRS systems are out there. I mean, there's got to be at least seven or eight at this point. Yeah, of course, it all started with, now it's called Lumenetix Core with the Mic by Apronoff company that sort of opened the floodgates.
We have our own CPT code 92229, which we've talked about a ton on this podcast and all our material at ainicare.com. And I think this is the topic for our future fireside chat, Scott, but the thing that's interesting is that the uptake still is probably not where I was expecting it, meaning it's lower. And when I ask my primary care colleagues on whether they're using this, whether they've heard of it, not a huge, not a ton of interest. I still think the bottlenecks and flow in clinic, I mean, we're asking primary care doctors and their staff to do more and more. And this has just not been shown to be a huge priority. I hope that changes. I think it may, as we talk about more about oculomics and cardiovascular screening and that type of thing.
So I think this is still setting the groundwork for that. But I think it's one more time, like we've discussed before, it is still a workflow ad. It's not a workflow subtract. And I think that's the single biggest hurdle. All these technologies have to get over. Yeah, definitely something we will talk about in future episodes. Scott, so what do you have for the knowledge about this week? Well, it's funny you brought up predictive AI because I was having that conversation this weekend with some people that are friends of mine. And they said, well, I don't understand how AI is going to be able to predict what we do in health care. And I said, well, it's interesting that you think that.
I mean, if you look at our large language models, those are predictive, right? What do they think you're going to want to learn and what's going to come next? I said, now, let's take this a little further. So for our knowledge by Rayon, I think, you know, we're going to break down that phrase of predictive AI. You know, I think we always ask, well, what is it and how is it going to change what we do every day? So let's kind of unpack this a little bit. So first, let's think about I'll call it the first pillar, which is awareness. And maybe even we'll throw in a little logic there. So right now, health care relies really heavily. In fact, I'd say we're primarily a reactive industry.
We don't do a lot of proactive medicine. We do reactive medicine. We wait for a patient condition to deteriorate or for some complication to manifest before we kind of get involved. And we are a reactive health care system. And this creates a lot of friction. It creates clinical stress. I mean, it creates emotional stress for the person who's going through it. It creates costly emergency type situations. Well, the theory is predictive AI is going to solve this by analyzing massive amounts of data, historical patient data, like, let's say, EHR, histories or lab trends or vitals to calculate the probability of future events before they happen. And in theory, that shifts medicine from a reactive state to a proactive state.
Now, there's one big, huge fundamental problem with this. You've got to have data. And we've talked about this many times right now. And without data, AI can't learn. And so then you're guessing, right? And guessing is dangerous. And so I still think, though, I look at this and I think that predictive AI is the future. I'm a firm believer that it is the future. But I think we need to have a better data source, a better realistic data source before we're really going to realize what predictive AI can do. So that brings us to our next part is kind of our next pillar, which is desire. Do we as patients, do we as providers, do we as staff think this will actually make life better?
You know, and I think it will. I mean, I think when we truly get to, let's call it, you know, predictive nirvana, will it remove this dread of, I don't know, I am just kind of caught off guard. I didn't expect this to happen. I thought it would be a few years before we saw retinopathy, to use your analogy from before. You know, for clinicians, the predictive AI, I think, I believe it's going to act like this quiet little copilot that I think we used that term before. Quiet little copilot that's going to say, hey, this is a high risk patient. We might want to watch for them. So instead of having to put out fires and when it comes in and it's already, you know, now we're already saying it needs to go to retina and have PRP or whatever it is that they're going to need to have.
Now, all of a sudden that we can empower health care teams and prevent them from being this reactive, burned out phase to more confident like, hey, we're going to do this proactive. We can proactively educate. We can proactively do that. Now that brings us to our third pillar. And, you know, I always kind of like talking about things in pillars. So, you know, because it's like this three legged stand in this case is two or three of them aren't going to make the stand, aren't going to make the table stand up or the chair stand up. So next we get that physical pillar, which is really kind of is, do we have the ability to create predictive AI? And the answer is yes, I think we do.
I don't think I think we're still in its infancy stage where AI will continually run in the background of everything we're doing. And then a patient's risk score will cross some threshold and the AI will pop up and say, hey, pay attention to this particular patient. The negative of it, as I said before, is there's going to have to be a lot of changes in the HR and a lot of data that's going to have to be released to develop this predictive AI. But I think we get closer every six months. I think new things happen and we get closer to do that. Now, I think the ultimate challenge, though, is not going to be awareness or logic or desire or ability.
I think it's going to be going back to that same theory we talked about is we as humans, whether it be patients or doctors, always want to go back to, well, this is what we did in the past. And it didn't work perfectly, but I don't want to change. So how do we reinforce, even if we get notified of predictive AI, saying, hey, this is likely going to happen? How do we prevent our own teams? How do we prevent patients from slipping back into, well, this is the way I've always done it, and ignoring these predictions. And we just go back to, yeah, we have the great technology. We just care not to pay attention to it. So hopefully we made predictive AI a little bit more familiar with the audience.
But I also brought up some things for us to think about and talk about as we're in meetings and we're in rooms and we're talking about, well, this predictive AI is going to do this. It's not just as simple as, can the technology do it? It's will we as a human race embrace it? Speaking of embracing. Yeah, a lot to unpack there and speaking of embracing and new things to talk about and new paradigms. I know you have a burning interest in this next topic. I do. I do. And it's about something we all care about. It's about money and getting paid and getting paid on time and reimbursements. And so how do you think AI and sort of closing with our fireside chat and sort of speaking and blue skying topics as we often do?
Your thoughts on payments, payment processing, reimbursement. It's a big topic, but what are the key highlights? I'm sure I'm going to make a lot of friends with this one. I still struggle every day, Rehan, with why our system has to be working with in the dark ages. I mean, our payment reimbursement system, come on. I mean, we're still doing the same things we did 25 years ago. Maybe it's not a paper chart we submit in the mail, but we send in some electronic resubmission and we wait 30 to 60 to 90 days to get paid. If we get paid, if we did everything right, and if we didn't do everything right, do we understand why the errors were and can we actually refile it?
And does it take another 30 to 60 to 90 days? And then once all that happens, then we have to send a bill out to a patient because we didn't know on the front side what a patient was actually going to owe. I think that's crazy. I mean, it is absolutely insane what we in the healthcare industry, not just eye care, but the entire healthcare industry. It is insane what we're tolerating. I just think it's crazy. In the world of technology today that we can't know, maybe not necessarily preemptively before the patient walks in the door because some things pop up and I get that. But there is no reason we don't have live time, transparency of what benefits and eligibility is.
To transport that into what are our costs going to be? What's our reimbursement going to be from insurance? What's the reimbursement? What's going to cost the patient? There's no reason that can't be live time. There's just no reason. Our financial institutions are live time. Why are insurance companies in the dark ages? Let me give you my thoughts because I think there is a superficial obvious reason. I think there's also a deeper reason, which I think is definitely more interesting. So I think the superficial obvious reason is that they're incentivized to create friction and ultimately, pavements being delayed and then fighting. Hope I will just quit.
Yeah, they hope we'll quit and sort of obvious. So I think that's the superficial reason. I think the deeper reason, which is maybe more interesting and maybe more interesting for our audience, is that when electricity was first invented, we would think that the productivity gains would be immediate. Wow, you have this new amazing technology. But in fact, what happened was that the factors had no idea how to implement this new found, newfangled technology. And so it was done piecemeal. It was not until you rethought, reimagined, redid the entire architecture, the whole system that you had huge productivity gains. So I think in the same way for a lot of industries, and I was talking about payments and insurance processing and all these things, you asked, why are we doing it this way?
Well, I think it will take a new upstart to do it in a way that's entirely reimagined, that is done instantaneously. And either one of these legacy insurance insurer companies will sort of start doing it. But oftentimes, it takes an upstart. It takes sort of a new innovator who reimagines how it should be done, who doesn't view AI as a newfangled technology that they're trying to retrofit in. But just building from the ground up. I know a guy. I said, I might know a guy that's working on that. So I think that's what is sort of needed in this. So it's going to be slow, slow. We're going to be asking these questions until one day it's, oh, someone else is doing it a different way.
So I think that's sort of my view on it. I'd be very curious to see what our audience thinks about it. I'm interested to hear your thoughts. I know this is a topic. And see if anybody's working on it. I mean, we can only complain about stuff. We complain about stuff, and I'm not a big complainer. I'm like, just if you're going to complain about it, figure out a way to fix it. So I want to know, is anybody working on it? Is anybody working on what are the challenges they have? Because if we start talking about the challenges to the audience, of all the people who listen to this, maybe somebody's go, oh, I never thought about that. Or maybe I have a solution to that.
Maybe if we all work together, we can solve this. And I love your point, this huge friction point that we've just endured and tolerated. Yeah. Until hopefully one day we won't have to. Yeah. Yeah. So to Scott's challenge, for those who are listening, if you're working on something, work on something like this, it's interesting. We would love to have you on our podcast, on our Innovators podcast, or even write an article for us. So we know there are a lot of people who are in this space. We'd love to hear from you. It is Scott S-C-O-T at aiandicare.ai. I am Rehan, R-E-H-A-N at aiandicare.ai. And you can find all of our material at aiandicare.com.
Thank you for listening. And we look forward to talking to you again next week. Have a good day, everybody.