Loading summary
Transcript17 lines
- [00:00]
A
This is Scott Becker with the Becker's Healthcare podcast. Thank you for tuning in today. Today we've got a brilliant physician leader and we're going to talk today about how artificial intelligence can drive meaningful results in revenue cycle management. We're joined today by Dr. Josh Galeras. Josh is the physician, the co founder and head of product development at Smarter Technologies. Josh, you take a moment and introduce yourself.
- [00:33]
B
Absolutely. Thank you so much for having me on. Definitely appreciate it. And as you said, I'm a physician. My co founder at Smartr DX and I trained together in internal medicine at Cornell and then I practice as a hospitalist at Columbia. I also have a software coding background and data background from bottom before medical school and was always sort of drawn to the power that technology could have in whatever I was doing, including medicine. And I used that experience to become at Columbia an NIH and DoD funded medical researcher doing data science and bioinformatics with electronic medical record data. I was also helping on a number of hospital operational initiatives and through that work I learned a lot about revenue cycle. As you know, every hospital is a little bit different and it's not something that physicians know basically anything about. I mean to be blunt, 99% of physicians probably know basically nothing about how hospitals actually get paid. And so the more I got exposed to, the more I was able to start imagining how technology and AI would be able to augment the quite manual work of revenue cycle. And so my co founder Michael Gao and I decided to start a company, SmartRDX in order to start bringing that the power of AI to healthcare revenue cycle. And we started by creating a product that we now call Pre Bill, which ingests clinical and administrative data from health systems, uses AI on the clinical data to try to understand what diagnoses and procedures were actively being treated during a patient's inpatient hospitalization during their stay in the hospital. And to compare what the AI generates as a list of possible diagnosis codes to what the hospital hospital's existing processes had cataloged in their review of that chart for billing purposes. And where we find diagnoses that weren't captured by that sort of manual process and would impact the hospital's reimbursement, their revenue or their quality metrics. Then we present those back to a human in the hospital for review to potentially capture. And the impact of that is something like $2.5 million per 10,000 discharges in new revenue that the hospital would otherwise have forgone for care that they are already provided and are completely justified in in billing the payers for. And you know, last year we drove almost half a billion dollars in revenue to providers that they otherwise would have missed out on. And 2026 is, is looking, looking great as well.
- [04:16]
A
No, it's really a remarkable background. I mean you, you've trained throughout the world. You also spent time teaching and working at New York Presbyterian. You're a technology person and a physician. It's really an amazing mix of what you do. Talk a little bit about Smarter Technologies was created in 2025 bringing together a handful of companies. Access Healthcare, Smarter DX, Thoughtful AI and PSYS Technologies. How would you describe the platform for this combined platform and what larger problem are you trying to solve together that's greater than the sum of its parts?
- [04:56]
B
Absolutely. You know, every year in this country we spend a hundred billion dollars on the administration of revenue cycle in healthcare, which is just insane. That's just like such a huge number. And I think that's a conservative estimate. Obviously I've heard much higher numbers for that. And as I think about why that's the case, actually my dad was a physician. His dad was too. But I have these pretty vivid memories of being maybe 7, 8 years old and following my dad around a hospital. And you know, this was early 90s and pre EMRs and, and what he was doing everywhere he went in the hospital was like walking and finding these, you know, thick manila folders full of patient details that told the story of what was happening for the patient. And he would, you know, sign his orders and write his notes in those folders. And the system at the time had to figure out how to get compensated for the care that they were providing. And the way to the best way to do it. You couldn't send that whole thick, maybe 100 page manila folder to the payer. You had to start by going through the information and sort of shrinking it down into something that was manageable and could be faxed. And that's where the process of, of medical coding sort of originates. And the DRG system, Diagnosis Related Group system that hospitals are largely reimbursed with today comes from just this process of trying to translate what's happening in clinical care into something that could fit on a single page and be fast at the time to the payer. And while we've largely digitized the system, we really haven't changed the underlying process. And one of the core things there, and this is what, you know, I think why we spend so much money on healthcare is it's largely manual. Right. We have humans going through all of These charts for every inpatient hospitalization and coming up with a catalog of the codes that were treated. And so the promise of smarter technologies is really about using AI to augment and improve the work that those humans are doing. And while AI is remarkable in many of its capabilities today, we think there are still an important need to have humans in the loop reviewing and doing a lot of that work. And so on the smarter DX side, as I said, we present our findings to a human at the health system. But not every health system wants their own humans to be the ones doing all that work. And so that's where one part of smart technologies comes in. Access Healthcare. It's sort of one of the largest and Most tech forward BPOs that's doing healthcare work in this country. And they have a great reputation for low cost and high quality. And we are working on integrating our AI products with their services team. Thoughtful AI is a company that's using agentic technology to help improve that efficiency as well, to make those humans even more effective. And then Pieces Technologies is an early stage company that's working forward in revenue cycle at the time physicians are actually writing their note. And the ultimate goal there is to integrate revenue intelligence into the physician experience and do it in a way that, that makes a physician experience with the revenue cycle better and more seamless and improves the physician experience in how you write notes in the technologies and AI capabilities that are available to doctors taking care of patients today. And altogether, our goal is ultimately to reduce the cost, cost of health care in this country by using AI to augment all of these manual processes.
- [09:38]
A
Thank you very, very much. When we talk about AI today in use cases, obviously there are a ton of different solutions being thrown at providers and health systems. Everybody wants to pilot something. Even doing a pilot for many, many companies or many health systems is a lot of work. It's not, it's not, you know, people say, oh, we're just going to do a pilot with you, it's going to be easy. But the lift in doing a pilot with somebody is significant. And so people have to choose where they're really going to double down, where they're going to do pilots, where there's going to be big wins. From your vantage point, which AI use cases are actually delivering our way today, you know, and what helps them to scale in a health system, what works, what doesn't work. Talk to us a little bit about that.
- [10:26]
B
Yeah, I was having lunch with a CIO at a very prominent health system earlier this week and one of the things he said was basically every time savings AI, like every time they've tried to do an AI initiative with a goal of just saving time as the value proposition, it's fair every time. And again, very large, very prominent health system and they, they're, they're doing a lot of, of these initiatives. And I thought that was, that was a very interesting comment. And, and it very much mirrors something that we've learned through experience at Smarter. You know, we piloted product with, with a large south east health system earlier or I guess last year. And it was, it was a registries product using AI to help humans who are going through medical records in order to abstract key information from those records for a variety of different registries, whether it's sepsis or an oncology registry or something like that. And what was fascinating to us was the AI got it right almost like 90% of the time. I think our final results before we shut it down were like 89% of the time. But it didn't save any time because the users of the system still had to go through and every single answer and sort of verify that it was right. And the work of reviewing that chart ended up just switching and replacing the work that was going into finding those answers in the first place. Right. And so, you know, I think that from a ROI perspective, which I think was a core part of your question, you really have to do a lot of work and get further than I think AI is really capable of today to generate really meaningful time savings in a way that systems trust at scale.
- [12:52]
A
Right?
- [12:52]
B
It's like, yeah, you can save time on this task or that task here or there, like exactly like you said, these sort of like point things. But to save enterprise time in a way that ultimately saves cost is just a hard value proposition. And I'll share one more sort of little tidbit here is with the Smarter DX pre bill solution, we end up finding more revenue opportunity than the cost of most of our health systems, entire CDI and coding teams. And so if we would have gone after a sort of time savings approach, even if we fully automated all of CDI and all of coding, we wouldn't have been able to generate as much value for the health system as we are today because our costs would be capped at what they're spending. And that's a really high bar because you have to fully automate all of those functions, which is just like really hard to do. And so, you know, I think that that's an important lesson that we've taken away is, you know, there Are there is role for augmenting humans and trying to improve their experience and save time. But that's not going to directly lead to roi. It's really about expanding what's possible within the health system and capturing more or, or doing a better job that is going to result in, in value. And I think that is where AI paired with humans actually makes the most impact on, on healthcare in the short term. Who knows what the long term looks like.
- [14:40]
A
But, but, but yeah, no, I think that's exactly right. And I love this, this concept that it's just technology is. It's this concept that technology plus people is going to be everything you're going to need both to make this work. Great. Talk a little bit about guardrails and innovation. Increasingly systems are having AI governance structures in place to try to make sure that people don't go off the rails and do all their jobs through AI Ended up in weird spots. How should companies, how should systems. How do you look at guardrails and innovation so things move faster mast but you still have humans in the loop as checkpoints. You know, how, how do you sort of put these two things together so one doesn't get way ahead of the other? Ended up in an area that's very different than expected.
- [15:28]
B
Yeah, that's, that's a, a great question and really important question and, and core to the, the smarter strategy really is to make sure that you, you're carefully attuned to what AI is and isn't capable of and that you have humans in the loop to make sure that where there is risk of AI as you say, going off the rails, there's a human there who's incentivized to make sure that's correct. And I think that's kind of the core thing is you, you know, it's very easy. I, you know, as a physician, I would somewhat ashamed to admit I used to like copy forward my notes, right. I take my note from yesterday on a patient and, and use it. It's very easy to sort of like miss the details. And so sometimes I would say two days in a row, you know, the patient came back from the OR yesterday, right. But they went to the OR once, right. And, and so that's a sort of example of where it's easy to skip over details. And so what you want to do is present information in a way that makes it hard to skip over the important or risky details that the AI is suggesting and not just give an easy button to moving on to the next case or, or, or accepting Those recommendations, I think that's pretty core part of having humans in the loop is, is that incentive structure.
- [17:15]
A
It's fascinating. It's fascinating and it's fascinating to talk to you because you've got a gifted way of explaining it and thinking about it. Take us to the next issue. So much of AI today and technology today is spent on pre authorizations, denials, revenue integrity. Where can AI most credibly cut avoidable denials and speed appeals in the next few years? And what traps are there is, it seems like this is an arms race between systems providers and essentially Medicare Advantage and big payers. How to sort of use AI to cut through some of this to level the playing field and keep payments moving?
- [17:59]
B
Yes, absolutely. I think that's a great question. And a arms race is a very good way to describe exactly what's happening, I think between providers and payers right now. I do think there's perception on both sides that each, that the other side is just trying to police them or trying to take advantage of of things. I don't think either side truly thinks of themselves as doing that. I think both sides are really trying to, trying to tell the right story. But it is quite complicated. And I think the problem promise of AI is really to simplify our ability to understand what's happening with patients and what's required in order to justify billing for care. I think that's kind of the core challenge in revenue cycle today and why it costs $100 million to administer revenue cycle today is it takes humans looking at every single chart, every single data point in order to accurately describe what's happening. And I think the real promise of AI is to take that process and augment those humans with better, faster understanding of what's going on. And I think that can play in a number of different places. So one is on the appeal letter generation side. At Smarter, we have an appeal letter writer that takes in a denial, looks through all the clinical data and pre writes an appeal letter for a nurse. And I think the core value proposition that provides is less on writing the appeal letter part. It's more on searching through the 30,000 data points that we get on average for an inpatient chart and identifying what the important information is in those in that data so that you can craft a strong appeal letter. It's really about pulling out and simplifying the process of trying to understand what was happening in that chart. I think that's also true even before discharge or while the patient's still being taken care of. A lot of the challenge around denials and appeals and certainly complex clinical denials is just understanding what the requirements that payers have in order to meet say inpatient status criteria are and to be able to very clearly articulate why a patient that you're taking care of meets those criteria. If I don't know what those criteria are, or it's a 47 page document that outlines what those criteria are, it's going to be very difficult for me to accurately describe to you how the patient I'm taking care of meets those criteria. But AI can understand that and very quickly translate it and make suggestions to the physician that that help describe that. And I think that sort of like simplification of complex clinical data in a way that humans can then validate and push forward is the promise that AI has in the short term to, to really take some costs out of the system. And we're quite hopeful that we'll be able to help. But I know others are working on it too and you know, for my children's sake, I'm, I'm really hopeful that we or someone successful at reducing the cost of care here.
- [22:07]
A
Let's hope so. It's become daunting. The cost for a family of four compared to the average income of a family of four are numbers that just don't work. And so if this is a way to help get more people focused on the right stuff and into clinical care versus so many people and time and hours and money spent on administration, this would be fantastic. Dr. Glarus, talk to us for a second. CMS is moving to faster, more transparent electric prior authorizations. How to change which AI investments may pay off first or does it change it?
- [22:42]
B
Yeah, I mean, I think this gets to what I was just talking about a little bit, which is in order to submit those electronic prior authorizations, you first sort of needed to know what patients need a prior authorization and what data is relevant to send to the payer in order to request that authorization. And I remain pretty convinced that the core task that we need AI to do is get better at understanding complex clinical data. And I'm really putting my money where my mouth is proverbially in the sense that at Smarter, we're really investing a ton of our engineering and data science resources in exactly this problem because I think it shows up in so many of the diff, so many different parts of revenue cycle. And the better we're able to do that, the better we're able to solve kind of all of these different problems. And so I think that is a core capability that we as a country in the healthcare system need is just that deep understanding of what's happening with patients and how to translate clinical data into the ontology of revenue cycle.
- [24:08]
A
Dr. Glares, let me ask you one follow up question. You've got this fantastic blend of deep sort of technology experience with deep physician and healthcare experience. Anything else that you'd like to share today that we haven't touched on from your perspective?
- [24:25]
B
You know what, what I will say is I while we have a lot of challenges in front of us and there's, there's a ton of risk, I think that I am pretty profoundly optimistic about what's possible over the short to medium term. AI is not a panacea. It will not solve all of our problems. And being really critical and thoughtful about how you use AI is very important, but it is a meaningful advance on how we do things. And the results I'm seeing come out of our team are extremely encouraging on a number of dimensions that really give me hope that these are surmountable problems. And even though we've been thinking about them and working on them as a system for decades now, I do sort of see a light at the end of the tunnel here where we can make a really big impact and shift these healthcare dollars from administration to patient care, to discovery to research, to things that meaningfully improve people's lives. And that's our focus. That's what I challenge the team to do literally every week. And I think we will see some real gains over, over the next few years. So I'm excited, hopeful and optimistic.
- [26:09]
A
Josh, thank you so much for joining us today on the Beckers Healthcare podcast. We want to thank you, we want to thank our podcast sponsor, Smarter Technologies. Just an absolute pleasure, Josh, to visit with you and talk about all the things that you're doing and the impact of AI on revenue cycle management. Thank you so much for joining us.