RISE Radio
Join the RISE team as we chat with industry leaders and explore ever-changing policies, regulations, and challenges faced by health care professionals responsible for quality and revenue, Medicare member acquisition and experience, and/or social determinants of health. Produced by RISE, the number one source for information on all things Medicare Advantage.
RISE Radio
Episode 30: Navigating the shift to prospective risk adjustment with hybrid AI
In this 20-minute episode of RISE Radio, Editorial Director Ilene MacDonald sits down with Dr. Matt Lambert, the former chief medical officer at Reveleer, to unpack the rapidly evolving landscape of Medicare Advantage risk adjustment. Together, they discuss the implications of the 2027 Advance Notice, new OIG compliance guidance, and the industry’s shift toward encounter-based, prospective risk adjustment.
Dr. Lambert explains how hybrid AI—combining generative AI with long‑standing clinical rules—can improve accuracy, reduce false positives, and streamline workflows for providers, payers, and risk adjustment teams. He also shares practical strategies for organizations beginning this transition, emphasizing partnership, user‑friendly workflows, and technology that meets clinicians at the point of care.
About Matt Lambert, M.D.
Dr. Matt Lambert, is the former chief medical officer at Reveleer and a leader in health care innovation, blending clinical expertise with technology, policy, and payment strategies. He previously served as CMO of Curation Health (acquired by Reveleer in 2024) and led digital transformation at Clinovations and The Advisory Board. He has been CMIO for New York City Health + Hospitals, RWJ Barnabas, and Johns Hopkins, overseeing Epic and Cerner implementations. An emergency medicine physician for 20+ years, he authored two books on health care reform. He earned his MD from Marshall University and trained at West Virginia University.
About Reveleer
Reveleer, a health care software and services company, uses machine learning and intelligent automation technology to empower health plans control over their quality improvement, risk adjustment, and member management programs. With one transformative solution, Reveleer allows plans to independently execute and manage provider outreach and data retrieval, coding, abstraction, member management, and reporting. Reveleer leverages proprietary technology, robust data sets, and subject matter expertise, so health plans can execute programs that deliver value and improved outcomes. Click here for Reveleer’s Guide to AI in Value-Based Care.
Hello, and welcome to the latest episode of RISE Radio. I'm Ilene MacDonald, the editorial director at RISE. Today we'll explore a prospective risk adjustment approach that uses hybrid AI to improve accuracy and reduce compliance risk. And to help me make sense of it all, we're joined by Dr. Matt Lambert, Chief Medical Officer at Reveleer a leader in health care innovation, blending clinical expertise with technology, policy, and payment strategies. Welcome, Dr. Lambert. Thanks for joining us today.
Dr. Matt Lambert:Hello, Ilene . Thank you so much for having me.
Ilene MacDonald:I wonder to start off, there has been so much regulatory news in the last, well, now a couple of weeks we've seen the 2027 Notice come out. Just the other day, OIG released specific compliance guide for Medicare Advantage. It's the first one in more than 25 years, an update. I'd love to know your thoughts about what's happening in the industry and whether you were surprised by anything.
Dr. Matt Lambert:Yeah, it's been a busy couple weeks for sure, hasn't it? You know, when the Advance Notice came out for 2027, I was both some parts surprised me and some didn't. Was I surprised that the rates were relatively flat going in? Not not really. You know, t they're using some new data, 2023 and 2024 fee for service data, which by law Medicare Advantage rates are tagged to. So I wasn't surprised by that. I also wasn't surprised by the reaction, quite frankly. You know, it's been it's been a rough year for some of the payers, especially United and some others. And whenever you see drops like that, I think there's just again more unknowns into this. So I wasn't surprised by the rate announcement, wasn't surprised by the reaction. I was surprised , and we talked about this a little bit in our preparation, wasn't with the unlinked CRRs, you know, the the retrospective reviews and submissions for conditions that aren't linked to a visit. They've been signaling this for quite some time, u that that they were moving this way. And we at Reveleer have been, actually when I think about my entire risk adjustment journey, whether it be back to old Clinovation days or the Curation Health days, or now the part of the Reveleer family, we've been creating, I'm sure it reflects my background as an emergency medicine physician and a CMIO, but we've been working from the point of care with all of this whenever we can. So we've been moving there just because we think it's the most you know, you're asking providers to spend their time and to exercise their clinical judgment on information. So it always makes it to have it best at the point of care. You capture it there, support it on a visit, supported with documentation. So I was kind of surprised and that for the retrospective parts of our customers, and for others in the that this is arriving. It's still an Advance Notice, right? It's not a final notice. And so this may change in between the end of March, early April when when these Final Notices tend to come out. And so I wouldn't be surprised if and coming to you today from DC, I wouldn't be surprised if there's a lot of discussions in town going on right now around some of this retrospective stuff. So and you can still submit them, they're just not going to count it towards the score. And CMS calculated that as a 1.5% decrease. If that gives you an idea of how of these conditions are still submitted unlinked to an encounter. So yes, both surprised and not surprised by the Advance Notice. And then right on the heels of that came, I'll call it an update I think from the OIG around audits. You know they came out really strong last year when they're gonna audit 540 plans. CMS was going to hire a couple thousand coders, they were going to change the the number of encounters that they reviewed, changed period where they would extrapolate extrapolate that. And then that that kind of stalled with the Humana case in Texa. So we heard the first word from CMS h some new guidance on this and some new rules for that. And they walked that back a little bit, right? They're gonna change the number of members that they review based on the size of the plan, which I think makes a lot of sense. If you're a small plan, extrapolating 200 patients is a disproportionately large piece of that. And then, just not to read too much into the tea leaves, but if they've softened their position on some of the OIG and given us more guidance on that, we might maybe see Advance Notice soften a bit as well. But this is also the same crew that said that we weren't even going do Advance Notices when this administration first came in. So your guess is as good as as mine. I try to make an educated one. But it certainly has been a lot of activity and is going to create another, I uncertain year. Last year was certainly an uncertain year. When you throw these new changes along with some of the still the ACA populations are still, what is that gonna look like? And we see a lot of our customers really hesitant to make a big move into ACA right now because of some of the unknowns. So I think you know you're gonna have to embrace change because that's the only way to move forward through this. But I think it is gonna be another unsettled year based on this this flurry of administrative activity the first part of the year.
Ilene MacDonald:And I know that things might change come April when the Final Notice comes out. But based on what the changes that they're proposing in the Advance Notice, what do you think is going to impact the providers, payers, the risk adjustment teams this year and in the future?
Dr. Matt Lambert:A Great, great point. Providers, I think again, they're going to look for this at the point of care. I think we're going to see all this move upstream. The point of care tools that are there, I think we're gonna see more demand for meet me where I am. Give me the best information you can give me, make sure it's in my workflow, in my electronic health record. And, if it's aligned with my clinical judgment, I'll accept it, add it as a visit diagnosis, it'll go out on a claim, link to that encounter. So I think the providers are going to, their expectations are going to ratchet up with this because if if the market has to move this way, then it needs to be as good as it can be for providers to adopt it. Payers, I think, are gonna have to either they're goimg tp have to look for partnerships, not only on this on the provider but they're going tp have to look for point of care tool companies to partner with to try and make these as tight as they can. I also think that this evidence trail is going to have to be cleaner and tighter than ever before as well. So I think providers are or excuse me, payers are gonna be very interested in how can we get this information to the point of care, but also how can we support where it came from and what clinical evidence was there and where was it, and demonstrate that easily in case in the case of an audit. Risk adjustment teams are going to be asked to even more efficient, I think. You know, that there's a lot of efficiency pressure on them already. I think they're going to be asked to do more with less, and that's going usually mean tech enablement, quite frankly, in this world. So I think any team is going to have to consider how do we use technology to especially the rote things, you know, the 80% of the time that is that are or you know, as I was taught in medical school, common things are common. You know, if those common things that you see across multiple members of a population, if the machine can do that, and leave the CDI team to do maybe the more nuanced pieces like cancer coding, or where, you know, is this an act of cancer, is this a history of cancer, is this a C code or is this a Z code? I think you're going to see tech solutions do the more routine and ask the CDI teams to do the ones that really require a little bit of human in the loop with that. Also maybe even some folks around the country might have to consider offshore positions, right? I mean, if we're getting more margin pressure and and we want some CDI teams, we might consider, and it's probably 50-50 in the market when we talk to folks about who's comfortable leveraging offshore and who's not. But I think those discussions are all going to come up in between now that final, you know, let's say whatever Friday afternoon they decide to release it.
Ilene MacDonald:That's right. At like 5 15.
Dr. Matt Lambert:Exactly.
Ilene MacDonald:Usually it's before a major holiday, right?
Dr. Matt Lambert:Yeah. So I think they're g oing to be all these and I'm sorry h interrupt you there, but you know, as we were preparing for this in the last couple weeks, we're just talking about when our customers or potential customers come to us with potential solutions. How are we going to address this? So we're actively working on solutions and strategy on the back end now as a as as the company Reveleer, and I'm sure everyone else is in the industry as well.
Ilene MacDonald:They better be. Are there any mistakes that you've seen with teams that have been relying too heavily on retrospective processes?
Dr. Matt Lambert:You know, I think mistakes might be a strong word. Too harsh, yeah. Yeah, yeah, because I think ht e best path to value, right? It was the path of least resistance for a long time was to do this on the back end, to have dedicated teams go through this information and find those those diagnoses that whether they're on a visit or not, and submit those in order to get you know credit for the complexity that you're managing. So I just thin that was path of least resistance and easiest operating model. There were most people worked in good faith with that. Famously there are some who did not. Uh, and I think because of it is why one of the drivers for moving this upstream and providers to validate these on an encounter, which is why it needs to be even, I think we need to be even better when we move upstream, right? We need to match the clinician. And again, I'm sorry for my doctor bias here, but if we're asking the clinicians to do this, that it needs we need to match their attention to detail with the information that we put forward.
Ilene MacDonald:I know that Revaleer has been focusing on processes for some time. So can we talk a little bit about how that perspective helps? What are the biggest workflow advantages both operationally and clinically?
Dr. Matt Lambert:We have been moving this way probably for the last three or four years and making some strategic acquisitions to do it. A lot of this expertise doesn't lie in one person or one company, so it had to bring a lot of different parts of the health care data ecosystem together to do it. You move it upstream because we can now, right? The APIs have matured. Interoperability is better, but not where I think anyone wants it to be. But we know when these patients are arriving, , we can pull that information forward, have a pre-visit team, look at the, you know, again, maybe let the more straightforward cases move directly to the provider at the point of care,, but have the have the CDI team look at those really complex cases, have the physician act on this at the point of, you know, at the point of care, and then that maybe a retro review becomes there's less value there, right? Maybe there's and one way to have less abrasion amongst physician groups is to minimize post-visit queries, right? So during that window where you know, saw the patient yesterday, I'm getting a query about to get some more robust documentation. If we can get that before the visit and at the time of the visit, then that could be a really big satisfier, I think, for the entire team to do that. And what we've been focused on. So a lot of the same some of the same data management, but moving it forward, looking at it prior to the visit and acting on it when uh at the point of care.
Ilene MacDonald:You mentioned earlier technology as an aspect in that now we're moving, having AI. It's a lot for many, it's a new territory. When we were talking in our prep call, you mentioned the hybrid AI. Can you talk about that in practical terms? What does that actually mean?
Dr. Matt Lambert:Yeah, I will speak about AI today because I don't know how fast it's going to improve tomorrow or next week. I mean, it's really it's really remarkable to do that. But think of you know CCDAs, think of paper documents on a member or a patient, and we're going to look at that information to look for chronic conditions that need to be recaptured or maybe a new condition that's evolved. So historically that was done with optical character recognition and natural language processing, so OCR NLP. Think of removing that because that was much harder to train. and anytime you were going to redo those models, you had to redo the entire thing, and it took months to months to do that. So think of that older technology removing and now having generative AI read those and produce this patient has type 2 diabetes with renal complications. This patient has atrial fibrillation. You know, of AI doing that and producing an output of the h and notes for things that we know the patient has and things that they might have. Now, and so those are much easier to train. We can just change some of the clinical indicators on the back end. We're we manage about 11,000 across the version 28 model now of things that we're looking for in a record. But we take that output, and then now we're going to run that through clinical rules that u my team and I have been developing for I don't know, a dozen years now, refining. We had to retire some of them, of course, with moving to from version 24 to 28, but ot a lot of those now map to the Part D model. So we just really kind of repurposed that or repointed a lot of those. So again, we're taking that gen AI output out of reams of data, and then we're going to pass that through rules that we've been, clinical rules that we've been building, validating, refining for a long time. And so it puts almost like a human in the logic, is what the team came up with. And I just love that term because you know, a human in a loop is still, I think for a lot of people working with AI is still something very important to them, and we can still facilitate that. But think of a human in the logic after the AI output before it even goes to a CDI team to do that. It's just a way to minimize false positives. If you want to lose provider confidence or if you want to get a lot of angry as a CMO, then when the when the platform produces false positives, that's the most frustrating thing. You know, you're asking me to spend my time to consider something that is falsely positive. And so the best way to do to minimize false positive is to take that Gen AI output, which is getting better every day, as I mentioned. But for now, we're going to run that through the clinical rules that we've been building and refining for years. And then there's even another opportunity to have a CDI at one of our customer partners review it as well before we put it to the provider.
Ilene MacDonald:If an organization hasn't yet made the shift to this perspective risk adjustment or incorporating this AI or hybrid I into their approach, where do you start? Is this a resource-heavy transition? I just can't imagine if you haven't done this, like it must feel overwhelming.
Dr. Matt Lambert:No, and once you get over that, once I'm sure there's some folks out there going through the stages of grief right now, right? But as I think and and the other thing for about the last couple of years that we've been hearing with either potential customers or current customers is partnership. You know, don't show me all the bells and whistles, show me what you have, but also show to me that you can work through any challenges or help us get from A to B. So the first thing I do with, I would think of if I was if I was a Medicare Advantage payer is I'd look for a partnership for what's my most engaged provider and that can again, that can be an engagement. That can also be are they on the right type of technology, i.e. an EMR, and is there a significant volume of patients where we want to try this, right? What what's the right size for that? Then I would find a point of care partner and and work on getting that information and from from the payer through this point of care tool to that provider group. So I so again, I think all about partnerships. I think all about finding that right partner, that right population. Know that what this looks like on day one is not going to be at all what it looks like on day 100. And build trust and then build scale from there. So same business rules apply. Let's do the basics well, get it right, and then expand from there. So, and and this can be done, done it multiple times. It just requires a lot of iteration, a lot of patience, and a lot of partnership. And sometimes that's the biggest challenge, right? The partnership piece. So I would really, before I did anything else, I would think who is our best provider partner that we're working with, and what technology can we we help them with? Because you know, worked on as a former CMIO, worked at health system, you know, IT shops for a long time. They all have a day job, they all have other priorities as well. So one of the things that we've learned at Reveleer is we need to make it as easy as we can for both the payer and the provider to share this information in a really effective way.
Ilene MacDonald:So if in this partnership, you're finding this provider that you want to work with in your health plan, what would you say for them as the first operational step once you've identified them to start preparing for this encounter-based world that we're now moving toward? Is there one takeaway that if someone was listening today that they should be thinking about?
Dr. Matt Lambert:I'm gonna inject my provider bias again here, I guess. I mean, I practiced for 20 years before before I got into the health care and technology space. Start focus on the workflow, user design. If you want the end users to act on then start there, make a experience where it's easy for them to do the tasks that we're asking them to do, despite all the other ones they have. And then work backwards from the point of care. A lot of time this comes the other way, and naturally so, especially if you're coming from the retrieval side. But if you this information to be acted upon, make that experience as easy as we can at the point of care. Make sure this information is clinically valid and make sure there's a good clean audit trail, right? This information was found on you know this CCDA from this date, from this line, went through this rule and was presented here. So those would be some of the guiding principles that I would say is start at the point of care where you want these conditions to be captured and then work backwards from there.
Ilene MacDonald:To those listening today, they're not sure where they want to begin, but they like what you've had to say, where can they find you and where can they learn more?
Dr. Matt Lambert:The team says the website is the best place for this. So Reveleer.com and specifically tab for the hybrid AI piece. There's a lot of information about that that came out when we we released on January 20th the announcement of the hybrid AI offering. And then also you can go from there to learn more about the point of care integrations as well.
Ilene MacDonald:Thank you, Dr. Lambert. I really appreciate your time today.
Dr. Matt Lambert:Thank you so much for having me, Ilene.