Your Best Defense Against AI-Driven Denials

Video

The healthcare industry is in the middle of a denial crisis — and chances are, your team is already feeling it. Payers are using AI-driven systems to flag, delay, and deny claims faster than ever, putting more pressure on revenue cycle teams to defend coding decisions, manage appeals, and protect revenue. Many organizations are responding with more technology: AI-assisted coding tools, automation, and workflow software.

But technology alone can’t defend a claim.

When a denial is challenged, audited, or appealed, it still comes down to having coders who can code accurately and confidently stand behind their decisions.

In this webinar, Corella Lumpkins and Shannon Cameron from AAPC's National Advisory Board explore why CPC-certified coders have become one of the most valuable assets in today’s payer environment — and why credentialing is becoming a critical part of a stronger denial prevention strategy.

You’ll learn how to:

  • Reduce preventable denials through stronger coding accuracy

  • Strengthen defensibility during audits, appeals, and reviews

  • Recognize where AI-assisted coding still requires credentialed oversight

  • Build a more proactive denial prevention strategy

Watch the webinar to learn why credentialed coders remain one of the strongest defenses against rising AI-driven denials.

Please note that this webinar has not been approved for CEUs.

Presented by

Full transcript

Great.

So we're so glad you're here. Today's conversation is about a challenge many revenue cycle coding compliance and audit teams are feeling right now. That payer decisions are becoming faster, more digital, and more data driven while provider organizations are still responsible for making sure every claim is accurate, supported, and defensible. Technology absolutely has a role to play, but the core question we're asking today is what when a claim is challenged, who can explain it, support it, and defend it? That's where certified coding professionals, audit auditors, and compliance leaders are becoming so important. This session will look at how organizations can move from a reactive denial management to a stronger denial defense strategy.

I'm excited to introduce our our speakers today. First, have Carilla Lumpkins, manager of coding compliance and provider education. She is and, Karela, keep me keep me honest if this is still applicable. She is at the Loudon Medical Group, which is an ACO in Northern Virginia and the DC suburbs. She has more than thirty five years of experience across health care revenue cycle, including auditing, billing, coding compliance, CDI, denial management, education, and practice management. And she's currently serving as a president of our national advisory board.

No. That's president-elect.

President-elect.

Our current president is the wonderful Rhonda Buchholz.

Thank you. Thank you, Karela, for president-elect. So future president.

Yes. Future.

We're also joined by Shannon Cameron. She is the chief operating officer for AFS of Harvard Medical Faculty Physicians. Shannon is an accomplished health care operations and revenue cycle leader with more than twenty five years of experience across multistate, multidisciplinary health organizations. She brings deep hands on knowledge of end to end revenue cycle, federal regulatory policy, managed care, revenue integrity, and finance across both physician and facility environments. And she's also serving as an officer on our national advisory board. So to get together, Corella and Shannon, they bring exactly the perspective that we need for this topic, coding compliance, audit, provider education, operations, and revenue cycle leadership.

So with that, we wanna know a little bit more about who we are talking to today. So we've got a poll that I'm gonna be bringing up right now.

Got this one right here.

So we wanna know where you believe your denial denial defense is breaking down today. Are you, having clear visibility into your denial root causes? Are there documentation gaps that are creating avoidable denials?

Is coding accuracy an issue you're seeing? Are appeals taking longer or not consistently successful?

Are you considering AI, but you don't know how to validate the output right now?

Or do you know the problem but don't have enough straight trained staff to fix it? And I apologize. I did only make this a one option quiz. So maybe there's a few of these that are that you're feeling. And if so, you can go ahead and chat about it. We'd love to hear if, like, more than one of these is something you're facing.

So I'll just give you a few more seconds to fill that out.

All of them.

Alright. I'm gonna go ahead and end that poll there, and I'll let you guys be able to see. So what we're seeing is the majority of you, appeals are taking too long or not consistently successful.

Documentation gaps are creating avoidable denials. You don't have clear visibility into what those root causes are.

Coding accuracy varies by coder. So it looks like there are problems all around, and this this is what we're seeing, widely.

So let's go ahead and end that. Alright. So with that, now, Carol and Shannon, I feel like you guys have some areas that you can speak to.

Absolutely. Thank you, Stephanie.

You know, this this this whole webinar and and what me and Shannon has come to do today is, just talk about the reality of what we're facing.

AI is coming. We cannot avoid it. We can't put our head in the sand. It's here.

It's already being widely utilized. If you don't have a tool currently in your practice or organization, then likely it's coming eventually, you know, eventually, budgets are being aligned to purchase these things. So it's here. So we gotta prepare for it.

You know? We can't be afraid of it, and we gotta embrace it. So this presentation, however, is, you know, some organizations and these vendors and these AI, software companies sometimes sell or oversell these AI tools.

And and so me and Shenandoah just talk today and just kinda go through that AI is just simply not the end all. You know, the human has to stay in the loop of all of this. And, you know, AI can generate outputs, but only the experienced professionals can really apply contextual judgment, clinical reasoning, compliance interpretation, appeal strategy. That only comes from humans in the loop. And so that's what we're kinda break down today. What do you think, Shannon?

I'll second that. I think, Karalee, you covered that very well. The only other thing, and I would emphasize is, you know, we hear auditing and auditing everyone thinks of coding. But in the AI world, with if you are an organization where you've implemented AI, you have to be looking at every part and piece that has been activated. We you can find that it can do a lot of great things that frees up the areas where we do need humans and we'll always need humans. But at the same time, there's a there's a mentality out there that you can trust AI and you absolutely have to be you have to have that human oversight and governance. But otherwise, Carrillo, I think you you covered it.

Yeah. I I really agree. So, so let's kinda peel this onion back a little bit and, dig into this a little bit.

So we'll drive jump into here's just what we're seeing in the market and the signals that we're seeing right now.

And as you can see, you know, denials are really you know, when we look at the external market, we're seeing this fifty percent site missing, inaccurate claim data, fifty four percent claim errors are increasing, eighty two percent reducing denials is a priority. It's a priority for everybody.

And so this is from our marketing or AAPC's marketing, should I say, of, you know, organizations of kind of the feedback that we're getting.

You know, that this is a true focus. And and so this really signaled to AAPC leadership and it really signaled to, where they're gonna be utilizing, like me and Shannon and board members and all the members that's out there and out here in the real world is, you know, targeting these denial. Sixty one percent rate denial audits impair scrutiny and are significant or critical. Fifty two percent say revenue protection will demand the most leadership retention, and seventy nine percent would protect denial prevention and revenue protection if budget to tighten. So it's not back office. It's something that we gotta put right in focus, but, you know, we can't rely on AI, you know. And, and and like Shannon, said earlier, you know, it's really gonna be a governance issue.

And this is where credential coders and experienced coders stay in the loop, stay in because this is not all gonna be just computer talking to computer.

So as we know, denial gonna create these chain reaction. And when we get these denials, what's gonna happen? You know, that's gonna bring our operations to a halt. You know, Shannon was just in, like, budget meetings, you know, earlier, you know, crunching numbers so she knows exactly, you know, how, you know, when money that we are anticipating do not come in, what that does to our operations.

So, go back, Stephanie. I I I don't think I wanted to you know, I wanted to just expand on these, you know, and really drive in that, the percentages of very or extremely concerned payers will pay and fifty percent of very or extremely concerned patients will pay.

So it's it's it's gonna be a chain reaction, as we go through this. Go to the next slide, Stephanie. I I think, we need to scale those decisions. I wanna get Shannon in here because, you know, I could keep going on and on and on.

You know? Shannon, what do you think? Because I I know we were just in budget stuff earlier, so I know you know how this affects the organization.

Well, you you've touched on it. One of my biggest things is one of the biggest operational changes occurring right now as what you've mentioned is the acceleration of payer review activity through automation. As they get more robust, they're using AI as well. So, you know, you have to make sure that whatever's happening you've to stay in front of it.

As Corella mentioned, I was in budget meetings almost twelve hours yesterday. It was brutal and it's because of where we are today. You know, planning for twenty twenty seven, we're trying to take in any and we used a lot of predictive, analytics that did come from our AI in the meetings yesterday. But, you know, whether organizations call it AI or predictive analytics, algorithm review, I mean, there's all these words floating around and they all pretty much are around the same thing.

It's intelligent automation.

Outcome the same. The basically payers can review claims at a scale so much faster than we've ever been able. I mean, feel like we're getting hit right and left. There, you know, if you're looking at your denials, you'll see a spike.

That was one of the things we looked at yesterday in our budget meetings was, well, what's going on? Why are we having more denials? It's because they're getting deeper and faster. And it doesn't mean they're correct.

Again, going back to what Carrilla said, you've got to have that human you've got to have, you know, yeah, you've got we have some AI automation, in our appeals and we we categorized our our, our denials by which ones can be automatically appealed. We've we're so many strategies we're looking at so that we can try to stay in front of And I could go on for days. But, you know, I know we've got a lot to talk about, but that's just like she said, put pulling those layers of the onion back, and that's just that's where we all are right now.

Absolutely. Absolutely. So, Stephanie, help us out with this Medicare Advantage example.

Yeah. So claims are getting challenged faster and more often, and, this is just showing that when met when Medicare Advantage went through, they had a big appeal case happen that four four four point one million of their requests were fully or partially denied, and eighty percent of their appeal appeals were overturned. So, basically sorry.

Basically, when Medicare Advantage is the denial denials create a burden of proof for providers. If the organization has strong documentation oh, I'm so sorry. Let me let me take advantage.

When I when when I looked at these statistics earlier, I knew it came from a specific source, and I had gotten it. So that's why I'm like, Stephanie, help me out here. But this just this this just shows, again, what we're trying to show, that these appeals are revealing the a a cause here. So if we are able to successfully deny if you know, and, you know, they're showing that they had over eighty percent of appeal denials was overturned in the end.

But Huge. That's huge.

I know that there were credentialed professionals in the loop there because that's what we're finding. We're becoming more on More on defense.

Than anything else.

And so that's where we're utilizing our skills the most. And I'm sure, that's what played a part here in these appeal denials and being overturned.

Okay, Stephanie.

So Go ahead.

Technology is not the whole answer. Yeah.

Definitely not.

We we gotta keep the human in. And and what we're seeing is gonna be a shift change, guys.

And, this is what I'm looking to take and elevate AAPC and help go through is that, it's gonna be a shift change because for us, not that our jobs are eliminated, but that our skills are gonna be used differently. We're gonna have tools for kinda the grunt work of these these things. But when we're talking specifically appeals and denials and when you're talking about the operation of your practice facility or organization being invested in these dollars that are, out there that we need to collect to keep our operations going. This cannot be completely left up to the robots.

Okay? So, you know, so we we gotta, make sure and that you know, I always say, you know, we're never gonna be able to automate health care completely because no matter how much of it's a bad word, there's gonna be diversity in health care. You know? Each patient is not the same.

We we're not gonna have there's no cookie cutter medicine. You know? So yeah. Go ahead, Shana.

No.

You're spot on.

I mean, there is you know, AI can suggest, but, again, the humans were you know, the humans are the ones that remain, you know, accountable. And so there is no cookie cutter.

Oh.

And then I I I I can't see how there ever will be. There's too many nuances. And patient's clinical presentation, that changes even by the same CPT code from one patient to the next. So much less adding a, you know, different, comorbidities and different things going on with the patient.

There's just no way.

Absolutely.

So then we come to this accountability gap because, you know, we gotta figure out, you know, who, you know, how this is gonna work, basically.

How is humans and AI is gonna coexist in this space?

So when we look at these denials and what they trigger off, and kinda how the if the workflow, how it is done, how this goes, AI is not gonna create every denial, but it can accelerate the discovery of it. So, you know, doing as many things prospectively ahead of time, anticipating denials, is your kind of our best defense. Remember, we gotta go on now kind of like in a a denial defense.

So it's almost like you, you know, as it it keeping the humans in is us anticipating the denials and already coming up with the counteractions from them.

Yeah.

I mean, technology cannot, you know, it cannot assume compliance liability, but clinical nuance is still you know, they still require that human judgment. I mean, one of the most dangerous misconceptions in health care right now is the belief that automation somehow removes the accountability, which as, you know, as we've stated it it absolutely does not. You know, it can suggest codes. Most of you know, a lot of people started seeing that. A lot of systems, we have the suggested codes.

You know, it can prioritize work queues. There's a lot of good stuff out there. I mean, identification of patterns and the predictability like we mentioned, but it cannot legally or ethically assume that responsibility for the coding decision itself. And so that's the part that we have to, you know, you have to ask those questions at the end of the day, you know, why was the code selected? You know, what modifier you know, why was the modifier applied? We, I talked about this at HealthCon in my session at HealthCon where, you know, we had erroneous modifiers that were added on to a claim and and it it it under it it escaped all of our governance checkpoints and auditing of the process before we went live with a particular new new system.

And, you know, it's those kinds of things, you know, and and are you gonna survive an audit? You know, at the end of the day, those are the things and those questions that have to be asked.

Absolutely. Ai is not gonna be able to determine all the time medical necessity just from the clinical nuances or the not all the time documentation sufficiency as far as, what's all all required for every payer everywhere all the time. You know, a payer plan will change up from yesterday to today.

You know? So, you know, so definitely, it's always gonna be a human in the loop as far as far as I'm concerned. AI can suggest where certified professionals can back up these decisions. And so it's kinda clear to us already what kinda AI can do and what the potential of it can do. It's gonna be even greater what it can do. But focusing more on what humans must do or what we own, it's never gonna replace clinical documentation.

You know? Again, the nuances of the diversity of health care is always gonna be there. When we are defending this, is, is the AI can we take the AI tool? Can we take the algorithm and put it on the the stand? No. We won't be able to do that. So there's always gonna have to be a human in the loop.

Be because, again, you know, as we know, we're in a industry in health care. The second most regulated industry in all of all industries is health care.

You know? So, so with all the regulations, rules, and laws, and changes, I can't ever see, AI being able to, just completely replace us.

So I'm looking at the comments. I'm seeing comments coming in. AI not determine medical necessity. I know Gina just said the same thing.

Yeah. You know, I'm seeing the comments have come in. You know? So, you know, so my, points are we can't be afraid of this.

The certified coder is always gonna be your last line of defense. So organizations is always gonna need this type of professional.

We're gonna be needed to detect, decide, defend, document. We're gonna be needed in all aspects, and we're gonna be needed for AI governance. You think that AI is going to regulate or govern another AI? No. It's gotta be some human governance over the AI tools as well, and that's where our skills is gonna be key. What do you think, Shannon?

I absolutely. I mean, coders are are there to detect those mismatches, and we see them all day long. Yet to see a system that doesn't have the mismatches, you know, between documentation and code selection. You know, the coders are needed the evaluation whether the payer policies.

That's the other piece of it. You know, that's where you start splitting off by, you know, the payer requirements. You know, actually support the service that's being billed or the procedure surgery that's being billed. You know, it's that validation of the modifier logic like I mentioned, you know, that's that's that's huge.

That's a whole presentation in itself, as far as just the automation and and, the edits that have to go into play. But again, it's just, you know, you're building off of what we call audit ready defense. And like Corella said, you know, you can't, you know, on the stand you can't say because AI did it that way. You know, it's the the humans are the responsible and also are the ones and it's not just the providers anymore.

You know, as coders, you know, we have to make sure that what's going out the door. If we're in a role, if it's your role to to to validate those things, you have to be sure that it's accurate.

The AI tool does not go to jail.

It doesn't wear stripes.

You know, it does not go to jail. So it is even greater. It's a greater importance for us.

Who's watching over the AI? So, you know, again, there's gotta be human oversight there.

That governance is gonna be there. You know, artificial intelligence right now, you know, legally, it's the wild, wild west right now because we don't have a lot of laws, in place for guidance there. So this is where we also, credentialed expertise, come into play as you're gonna be there to kinda, know the rules, understand everything, and you we're gonna be the ones that's gonna have to put governance governance on our organizations and how we use these tools because the AI tool, the vendor is not gonna gonna control themselves. They're keep going, going, going, and going, and promising more, more, and more.

So it is gonna be up to the credentialed professionals for the accuracy, the integrity, the, you know, just to be ready for these audits and to validate them. We're always gonna have to have a validation process for everything with these AI tools.

And and one other thing, you know, this is a term that I've been hearing lately is there's another emerging risk kind of, issue and it's what we call is AI hallucination.

And and what it is is in health care, it it becomes and it's very dangerous because what happens is the document the documentation, often sounds like it's clinically appropriate even when it's unsupported, you know, and it's very obvious to the human eyes when when something is off. And, you know, for instance, you know, AI may infer a higher risk level. Of course, we talked about that with the suggested codes. It can also introduce diagnoses that were never truly assessed. Carole, I don't know if you remember last summer when we did some testing, we were asked to test, you know, some some software and there was a lot of inference that went on with some diagnosis.

Absolutely.

Absolutely. And and and and even in my own organization, we allow our providers to choose their own AI tools, so we have several. But we found out that we had to put things such as, like, word counts in.

If the conversation the actual conversation was two thousand words between the provider and the patient, how are we getting a five thousand word document following the visit? You know what I mean? So we actually had to look at word count and things of that nature. So the hallucination is crazy, Shane.

Yeah. And once that supported information enters the legal medical record, you know, it creates a downstream, compliance issue that can be a nightmare for an organization if they're not on top of it.

Absolutely. And the AI will take some assumptions if you are not careful. So Nope.

That that's a great point. It does.

It does take assumptions. So, and it's only as good as the tool that's being used. You know? So you gotta be at the table and even ask how often is this tool being updated? How often is this tool being if it's not being revised and updated on a regular basis, you know, it's gonna be outdated as soon as you purchase it, as soon as you purchase AI tool. You know?

So It's changing at the speed of light.

You're absolutely right.

Absolutely. Absolutely.

Okay, Stephanie. So this is what our defense looks.

And, and and, you know, as, what the process is. If you are out there in the industry and you're working, denials, you know, it's like putting on your armor every day and going out for the good battle.

You know? And and and we always have payer plans that join our discussions. I always don't wanna seem like I'm beating up the payer plans, but because I know they're trying to be good stewards of their members' dollars of health care costs and and and keep those reined in. But if you're working for a provider organization, you know, you know, every dime counts. You know, if unless you're working for a provider with VIP clients that's paying cash at the door.

If you're working with any organization that has to deal with insurance, then we have to battle. And so these tools, again, can be helpful. You know, an appeal through through the lens of these AI tools, you know, it's just gonna go through the analytics that it's programmed into.

It's still gonna take that human, that credential coder, that's gonna strengthen each of these links, because we're gonna know where the AI dropped the ball at and how to address it. You know? And, again and personalize it because, again, in each as you know, doing appeals, you know, there are nuances with appeals. You may be appealing the same thing, but, again, each appeal letter has to be personalized to that patient, you know, and addressed as individualized. And so I I see where there's always gonna have to have that human touch.

So we already know the workforce is already reworking. You know? We know that ninety percent of claim denials are reworked at least. Forty three percent report claims teams are understaffed.

Everybody's understaffed, for the most part right now. And staffing shortages, that's creating a issue, because, of course, that's gonna slow claim submission and denial resubmission. And to be honest, that's where a lot of organizations are, looking to. I haven't seen AI come in and replace a whole lot of humans yet.

I'm not saying that we haven't seen layoffs. As you know, some of these risk adjustment companies, you know, contract coders and lay them off periodically and then till they pick up new contracts. There's a lot going on in that space, but, we still need it. We're we're still out there.

We still need to be out there because we already know, everybody has to rework their denials. We can't just let denials fly if they can be defended.

Yeah. And rework claims is the is by and large one of the the cost most costly the the the costly areas of revenue cycle. And that's where, you know, the AI can be extremely helpful. Again, by, you know, by looking at the analytics, taking a look at it and minimizing, you know, the redundancy of the simple of the simple appeals or the simple or the simple write offs.

Things that can you know, I always ask groups for those of you guys that work in in the revenue cycle on the AR side, why in the world would you have a human writing off sequestration? Sequestration, there's nothing you can do about it. You should have that, you know, if you have the capability, that's where AI can come into play. Write that stuff off so that you free up, you know, your human to actually work in those areas.

And and that's where we, you know, we always hear that that term, you know, work smarter, not harder. That's where using AI to take care of those redundancies and those simple tasks so that you can free up that time, you know, for for for the human, factor.

Absolutely.

Absolutely.

So we gotta have a winning model. And and that winning model is we, you know, we need these AI tools. Again, you know, you know, it does help. It's not, anybody that's out there using the AI tool right now, you can't say it it doesn't help.

You can't say it didn't do things, you know, faster or give you a starting point or or help or assist in in some type of way. So they're gonna be useful. It's how we use them, and that's what's gonna be the winning model. It's making sure that we are utilizing the tools.

We're gonna utilize the tools, but making sure that we keep that human in the loop, that credentialed experience, that expert, that expertise is what is gonna be upscaled. The AI is gonna be able to do the grunt work. The AI is gonna be able to do the, you know, the filing or the processing of things, if you will.

But it's the decision making piece, the critical thinking piece that's gonna always rely on that credentialed human in the loop.

So have a plan. And anybody that's dealing in this space, we usually do. So this, doesn't, isn't foreign to anybody of of just having kinda like a checklist for which kind of things that you wanna go through and what you wanna have in place to prepare. Because remember, we're on defense now.

So, typically, not only, are we fighting denials, you know, these these payment retractions and and coming back years later, increasing more and more. And, and as you know, CMS and our government has kinda doubled down on oversight of fraud, waste, and abuse, in health care revenue, you know, reimbursement space. So everything is being looked at with an evil eye right now.

So, you know, so we definitely want to make sure that we're have some kind of plan of action of things that's being done before you submit your claims in, that you have a process in place.

After something is denied, you know, you gotta find out what the denial cause or reasons was and attack it from there and then have those ongoing controls.

Audit. Audit. Audit.

Make sure that you're always going through. Make sure you're doing review. Again, things can change from one day, one week, you know, from another. So we, that's a job in itself sometimes and just keeping up on policy changes.

And and so, you know, so those are things that if, you know, the right tools could assist us with. But, again, we gotta have a human. The human coordinates all of this. This doesn't all kinda happen serendipitously. You know what I mean? So a human is there, always in the loop of all of these things.

And, Carol and Shannon, I'd love if you could maybe give an example of what this can actually look like.

I'd be happy to. I can't emphasize enough and very well said, Cruella. And I could not emphasize enough this standardization of your appeal package and what that looks like. So, you know, some some organizations they have, you know, fancy tools, fancy reporting, but you don't even, you know, you can take your own raw data and create your own standard, package.

You know, that's an area that if you know how to do a pivot table and drop those things in, have your buckets ready, but also look at it on so to give you a visual, look at everything on a twelve month rolling, you know, by by reason code. I mean, excuse me, by root cause, you know, so we bucket the denials. We bucket them into categories And look at that, you see the spikes. And then again, you take a look at that and figure out if you do have AI or the, you know, you have access to AI, use, you know, use that data.

Again, as I've mentioned earlier, not to kind of repeat myself, but that's where it comes in. But you you take a look at that with from a reporting standpoint, you have to know and you have to share that information and then it passes over to your CDI teams. But having that root cause understood and bucketed the way that you had and you know, we have a team approach. We bucket it in different people.

We have the the denials route automatically to a different team of people. So we have our our coding, our senior coders that, you know, handle the medical necessity and the coding denials. And then of course we have some that route over to credentialing and then the prior auths and things like that from the front end. You know, at one point, our denial, sixty percent of our denial, sixty, not about sixteen, sixty, which is mind boggling, were were patient access denials.

That had I not had this broken out system, I wouldn't have have understood that. We don't have a lot of control in that area in the revenue cycle part of the on that that side of the fence, but I could share the information with leadership and those patient access, directors who do have ability to to make change. So having that kinda laid out is is super important, and I think we're we're going into an area where we're gonna talk about that a little bit more.

Yes.

So metrics. Metrics is exactly what Shannon was just talking about, utilizing that information to put up your defense. Again, bucketing, I love bucketing the denials and having, and working that approach that way. But there are some questions you can ask yourself. You know, you is your first pass claim claim rate good? You know? That's looking at is the claim clean to begin with.

You know? Again, I like to say that I like to put a lot as much into the preparation and kinda work ahead of instead of always working behind. You know, if you know that you're always working, you know, everything looking at everything from the at the end of the output, to kinda start your focus and look at everything on the beginning of output.

Is is coding quality improving? You know, you gotta have someone looking at that. Always looking at day the day days from the denial to the resubmission.

You know, I always say that payer plans are smart, and we have to be just as smart on the payer side the provider side as well. Payers, I've worked at payer organizations, and then payer organizations, we have dashboards.

You know, I worked on the payer side where, as soon as I took fielded a a call back in the day, we used to get take calls from our provider, practices.

Soon even then, if a practice called or a patient or provider called, I could pull up. I already know, how many days and, you know, how long it's taking you to bail us. I already know how many claims, last year were denied, how many claims you resubmitted, how many claims, you tried to appeal, how many appeal doubts, how successful you are on your appeals, how many times we asked you for a medical record request and they weren't sent back? We already have all those stats, all those statistics right there just as soon as we make contact. So the health plans are already forming an opinion. The AI tools is gonna take that data and form an opinion, and that's where a human's gonna come into play, where we got to, break up the opinion, and answer all these questions when we're dealing with what's going on in our organizations.

Yeah. And again, I can't emphasize metrics enough. It's critical to know your numbers. Spot on, Cruella.

I mean, one of the biggest mistakes in denial management is focusing only gross denial volume without understanding and underlying, you know, what's going on. Like you said, having an understanding of what's a second pass upon appeal. What's your denial rate then? You know, make sure you know the the before and after.

And the are you doing anything with that information? So if it is getting corrected and getting paid upon appeal and it's something that you you you you change with maybe a senior manager looked at, a coding manager looked at it and changed the coding based off the documentation. Did you change that? Did you change the behavior?

You gotta use your metrics to make that change. And and like like Cruella mentioned earlier, they're going at rapid speed. We have to stay in front, you know, we have to stay on the front end of that. And again, the low clean claim rates that right there, that's a number that, you know, if you have your KPIs and you know your number, right there is telling you there's there's something that there's some instability somewhere.

Absolutely.

Let's talk about how to build the system.

Oh, yeah. We we definitely gotta build the system.

You gotta have go through your processes at your organization and have a plan in place. Think about things, you know. When we started getting into the day to day, you know, grind of of of keeping the our practices and operations going, you know, there's gonna have to be some discussion on bringing together all parties to kind of work together to kind of build a system. It takes a team. You know that it takes a village to raise a child. It takes a village to get a claim paid.

You know? So Right. You gotta have that same mentality. So it's only gonna be as strong as the people.

And and so we gotta have good, we gotta have good folks, and we gotta continue to learn. That's the the thing about our industry the most is you cannot stop learning.

You can never stop learning. This is a continual learning process.

You know? So so AI is gonna help us, but, we're still gonna need that human, educated, credentialed professional that's gonna be able to come in and answer, you know, what's missing from this?

You know? Oh, you know, that could have been something that the AI tool missed.

Again, medical necessity is individualized to the patient. That payer plan rule, AI might not have caught that there's some caveat to it.

Did that decision survive an audit and appeal? This is where the human comes in and that knowledge base is needed.

So we wanna definitely use our metrics.

Metrics, metrics, metrics, anybody dealing with denials or dealing in a CBO or billing office or responsible for any type of accounts receivables. We need we need the the the the the knowledge is power.

If we know what we're doing, then we know what to fix. Then we know where our human, you know, strengths are. We know where the AI can help us, and we know where we need to step in at. So metrics is gonna decide where AI belongs in your revenue cycle, and it's gonna be unique to every organization.

Again, nothing cookie cutter here. You know, you know, I was talking to a AI vendor the other day, he was telling me how, you know, he was making something and he's they wanted me to look at it. I took a look at it and he was making something that they wanted to release across the board. And when I looked at it, was like, that's really something that's an option and I think that you could keep it as an option for your providers.

But to turn it on to all to your entire you know, that's something that's very unique. It was a way that the doctor was did the templating in their, natural language based processing program, that was unique. It was unique for that specialty, and it was unique for that provider. And, you know, and they thought it was neat.

I thought it was neat too, but not nothing that we could turn on for all providers to use. Again, the vendors don't really realize how individualized things are. They don't know the provider and the physician, you know, have different preferences. One doc can love it, the other doc can hate it.

You know, there's a, you know, there's a whole provider for you know, I can write a book on just the physician personalities and provider personalities of of utilizing. And we see it. It comes out with the use of these tools. So metrics is gonna help.

That metrics show, all the information that we need. It's gonna be able to organize it for us, score it, prioritize it. But humans still gotta govern it. Humans still gotta have oversight of it.

You know? Final judgment is still gotta lie with the human. Human must stay in the loop, and this is where our credential coders come into play.

AI needs expert validation before it can be trusted. So, Stephanie, explain to me, what AAPC did with their work with the Ambiance Healthcare.

Yeah. AAPC worked with Ambiance Healthcare. They did an AI validation of this tool. So Ambiance is one of those listening tools that providers can use when, seeing a patient, And they were able to help support and validate the coding that this tool is do doing, by going through, you know, all of the behind the scenes AI.

And then with that validation, they were able to help improve it to be to get to an over ninety five percent accuracy across seven specialties. So they implemented an audit and reiterate process, that really helped improve its accuracy. And so this is just showing that this is just, you know, backing that while these AI tools are, you know, useful in helping us be more efficient and faster, we really need someone to help validate the accuracy, validate those codes, and this is how APC is able to do it. And, you know, what we're recommending for others to do is to really be able to go in and validate that code.

It it it totally proves that the AI can't govern itself.

Yes. So it would the AI, you know, that a human still was able to come in and find these errors to make the to get the the to improve the AI. So we still need human expert oversight to improve the AI performance.

Yes. Absolutely.

And then the next part of that is just how we can build continue to build that defense system and what that specifically looks like with AAPC's credentials. So all of the different processes that you guys been have been addressing is not only just CPC holders that can be the ones that can help defend that case. We can look at CRC, CPMA, and then our upcoming CPCM. And so I'm I'm wondering if you guys could talk to different, credentials and how that makes a difference in the education that a credentialed coder would need.

Oh, absolutely, Shannon.

We we wanna take this, all these different Yeah.

I mean, we've, I mean, a multispecialty, organization. So, of course, the right the right coding, you know, the right coding specialty for the right you know, for instance, we have we have OBGYN, we have anesthesia, interventional pain management. And those are those those those specificity areas where those coders knowing their criteria deeper than just a coder who has just a CPC. But then we also have, you know, when you start talking about the CPCM, those are the that's that new credential which is fantastic.

We have several people on our NAB team that have that have taken that and and passed that. And, it's a it's another layer of defensibility, you know, as far as the deep level of of, you know, oversight and understanding. So those those credentials, you know, again, having the right experience in the right areas. You know, we have we also have, our CPBs for our billing, you know, staff several years ago.

I took my group through.

We we did a an outside, of the organization, class and and got a lot of people in in, you know, certified through that. And I've seen leaps and bounds that the benefits of that, you know, that certification. Anywhere that we can build the revenue defense around the certifications because it's just deeper, deeper levels of education. It's just not just a, it's not just a certification to throw on the back of your name. It means something and there's a these organizations have so many, you know, depending on, you know, your specialties and and what you offer your patients. Then there's different, you know, there's so many specialties now that, the certification really deep dives into that level of protection.

Absolutely. And, all of these credentials, particularly the core ones, CPC, COC, CRC, all the green, core ones, those are our boots on the ground.

You know, that new credential CPCM, you know, that's they have kinda overview of everything. That's not gonna be a a credential that everybody gets because you're really saying it's beyond we're already saying that we have a professional level or expert level with your core credentials. Now the CPCM is almost a mastery level, if you will. So think of it as the next level up. But I also am, very realistic to let people know that everybody can't get that, because that actually takes that you have your hands on kinda everything from beginning to end on the revenue cycle.

But it's a wonderful credential if you do. I can't wait to take it. I'm gonna take it. So we'll see. You know? But it is very difficult and very challenging, so we'll see.

So the future of the denial defense is the system, not the tool. And this what we hope that we were able to convey today that, we must have human oversight, that there must be a human in the loop, if you will.

Your credentials are actually gonna become more valuable in this AI environment, you know, because, it's gonna be needed that you have demonstrated some type of expert or mastery level, to this, that kinda makes us, you know, have a little bit more insight, if you will, and know the day to day, every day things that maybe they are tool, is gonna be missed. So in five years, what is it gonna be? What's what's the future gonna look like? You know? Is it is it the organization with the best AI? Is that the one that's gonna be winning, or is it gonna be the organization with the strongest governance around their AI and how they I love that.

So I love that.

That's a fantastic fantastic point. I mean, the future revenue cycle, in organizations will not be fully manual and it will not be fully automated. It's going to be a combination of govern you know, a hybrid model. And that's the I think that's the biggest takeaway for me, for making sure everyone does understand that. And like Carilla said early on, don't be afraid of it. It's here. Embrace it.

You know, it's helping us scale and detect, prioritize and and identify patterns. Those are the things that are going to help us and just make us better. You know, credentialed professionals will remain responsible for the interpretation and our compliance oversight and, you know, our final accountability.

And, you know, defensibility exists, you know, exists upstream and audit feedback is going to continuously improve performance. And we have to think of it that way. It's just like in the early days, You know, a lot of coders were afraid. You know, I I remember being afraid of audits. Then I was I was happy to get those because it just made me better. And you have to think of AI that same way is is how can this make me, you know, me be a better professional?

Sure. Sure.

Alright. That that wraps us up. And before we do wrap up, we did just have a a final poll to kinda understand, here. Let me get this then.

Yeah. If AI is already touching your documentation, coding claims are down workflows, how are you currently validating that? Are you, do you have a formal audit process? I mean, what we like to think is that the first way to understand where, weaknesses are is to have a formal audit. Are you already spot checking outputs?

Are you relying on vendor reported accuracy? Are you building a governance process?

Are you not sure where AI is being used in your current organization?

Or are you not using AI yet, but you're embracing it or, you know, bracing for it and trying to get prepared for it? And it looks like the majority, that's where you are, is that it's not currently affecting you yet, but you know it's coming. So it's awesome to see that you guys are here already trying to do the work to understand where your coders are gonna be positioned when those tools are gonna be implemented. And that's exactly why we're doing webinars like this, to help everyone understand how to get ready for it.

And so I did just wanna do a soft plug for the CPCM. We do currently have the beta test open right now. So if you are someone that has been in health care for at least, has been a coder for at least five years, we are looking for anyone that wants to take that test right now. It's free, and it would be a way for you to kinda test your knowledge and see where where you're at and also validate your expertise.

And then we also have a an, event coming up called Docicon, in August.

And, Corral or Shannon, are either of you involved with that one at all?

Yes. I am. Can you speak with at DocuCon. Yeah. I'll be I'll be actually covering, I'm trying to remember. You put me on the spot. I'm trying to remember the topic, but I it's a it's an AI topic.

Really? Much in line with what we talked about today on a little bit deeper level.

I'm trying to remember the title of it, but I I'd have to go Yeah.

For that. But yeah. But I'm excited.

You can find you can learn more at a a p c dot com slash doc u con.

And let me plug in RiskCon. Alright? Before that Yeah. Monday's on. I'll be speaking at RiskCon.

So for our risk adjustment professionals or anyone in that value based care space, check us out. I think that's coming up June, next month, the twenty third, twenty fourth, I believe, next next month. And I'll be talking about AI tools. Isn't that amazing?

AI tools that we're using in this space specific to value based care, quality, and risk adjustments, for these, you know, agreements are changing, so we have to include that in there. And so, my session is about not recreating a wheel, taking what you have right now in your organization and, coming up with a plan because, again, it takes a team to get this payment.

So I love it.

And then we can go ahead. We have five more minutes. So we will take some of these questions that we have in here. Do you guys know of any AI software that's, handling denial appeals?

I know we talked about how the coder should be involved here, but is there any AI software that's doing Absolutely.

Let me jump right in. We are piloting, AI tool right now, with Waystar. Waystar is in that space. But, again, they're not the only vendors, but, check them out. And a lot of the vendors also, have demos and trials. So any tool that your organization is considering, once you identify what function you want the AI tool to do because there's no one tool for everything.

No. And, Trizetto is another, vendor that we've utilized, that, we were actually their guinea pig of creation of the auto appeal, with their arm product, and we had some great success with that. And, again, that's starting small, you know, again, we didn't have time to really go into to to granular details. But anytime if you're in a position where you're part of the decision making or implementation, you know, start small.

And that was that was one of our first steps of implementing, you know, we found the low hanging fruit, the the easiest simplest, you know, denial that could be auto appealed that didn't need human eye. A lot of them do need human eye, these were one of, you know, some of those that, and it we had great success with that. We also built in some, automatic write offs. Again, sequestration is a great example of that.

Why in the world would you have a human touch that and, you know, save your humans for the things that need, you know, a brain to think about and deep dive into the so yeah. But Zetto is is one of the ones and, again, I have no attachment or affiliation with TriZetto, but we do know that we have some good success. So k.

Yeah. Let let's make sure we say that. We do not have any contractual obligations to any of these AI vendor companies.

That's good. That's good.

Tracy asked if AAPC has plans to assist in helping organizations understand the role of the coder, auditor in the work play workplace with AI becoming more prevalent.

Yeah. Many are worried that AI is being seen as the cheaper option, and that paying humans is gonna become, you know, a budget cut conversation.

Well, what I can tell you already from going through implementations in my own organization is we start off that way. Senior managers and your c suite when they are looking at these tools to bring them on, you know, I'm not gonna lie and say we're not looking at, hey. If we can get the tool to do this, maybe we won't have to, you know, we'll have to have so many FTEs or so many so much staff. Maybe there's some kind of cost savings there.

That's what we have went into it with the thought process. But I'm telling you from being in the trenches down here in the dirt, that has not been the case. We have not lost a single staff person. And in in in some cases, in some areas, we've had to increase staff, for the use of some of these AI tools.

Now I'm not gonna say that that's gonna remain the same way all the time, but what I will say is that you have not seen any huge, okay. They laid off everyone at one organization specifically because the AI tool had taken over the whole function. We have not seen that yet. And, so I don't think that's, I don't I don't see that.

I don't see that in our space. There's gonna be places. There's gonna be things that can be automated as Shannon has found out. I thought that was a great use of how they used, their AI tools.

But she knows. You know, we still gotta we still gotta have oversight. We still gotta look at it. And, again, we still gotta have look who we gotta police the AI.

There's nobody policing the AI.

Yeah. And I think our, presentations like this are helping defend and provide you, thus audience, with the tools to go and defend the need for humans in the loop, and where these systems are breaking down, what questions a coder can be asking that can help, understand where those systems are breaking down. And it all goes back to the metrics as well.

And the number one question, what's our governance plan? That so many organizations are not thinking about the governance piece of it.

And you can't rely on your AI, the AI vendor. You know, a lot of them come in and, you know, they're selling you some, you know, program and it all looks good. But none of that stuff is cookie cutter. I'm telling you, you have to individualize it to your practice and organization.

So, so they don't but they don't control themselves. You you're gonna have to have, something in place at your practice or organization to govern that use of that tool of what they can do. Remember, a lot of these are IT type people. You know?

And before, when they first do these deals, a lot of times, they just hook up with the IT staff. So, you know, so you're like, oh, wait a minute. Did compliance at the table? Is the AR at the table?

Is the, you know, risk management is you know, everybody sees what is where's the provider at?

Oh, the provider definitely gotta be at the table because they gotta understand the NOS too.

So, so, again, I just don't see it. I see a a human always in that loop.

Agree.

Alright. Well, that brings us to the top of our time. Shannon and Carella, thank you so much for your expertise in this area, and thank you everyone for attending. If you're leaving today unsure where your denial defense is weakest, we recommend that you would start with an audit. There will be a question at the end of this. If you'd like to learn more about AAPC's audit services, go ahead and indicate that. If you are validating an AI tool, we can help you with that as well.

And if your team is just struggling to keep up, or even start this process, like we've been saying this whole presentation, start with education, start with credentialing, and we can help build a system around this technology and help you address these issues that we're seeing.

Alright. Thank great holiday weekend.

Yes. Have a great holiday weekend. Thanks, everyone. This will con conclude our presentation.

Thank you.

Codify Home Pricing

Validate AI before launch

Ensure your AI delivers accurate, payer-aligned results with real-world validation that reduces risk, improves performance, and accelerates adoption.

AI Impelementation OG 1200x628-1

Intro to AI Concepts for Medical Billing and Coding Course

Learn how AI tools are revolutionizing billing and coding processes. Understand emerging trends, improve productivity, and elevate your skillset.