Maximize AI's Potential in Your Revenue Cycle

Today, we are thrilled to in, introduce Stephanie Scott, who will be your presenter today. She has over twenty five years experience in the health care industry. She's worked in a variety of settings, including hospital, long term care, large multispecialty physician practices, and in EHR software design and development. Stephanie has extensive experience in the fields of auditing and compliance. And currently, Stephanie serves as vice president for AAPC's audit services where she manages an international team of coders and auditors and is responsible for the old overall project performance and client satisfaction for AAPC customers.

And with that introduction, let's go ahead and turn the time over to Stephanie.

Thank you, Cordell. I am so excited for this webinar today. Thank you so much for taking time out of your busy schedules to join us.

AI is top of mind for everyone. We see it everywhere we go now, and it is moving forward very fast.

We're seeing an impact in AI and health care already. There's lots of early adopters. There's a lot of information out there. And so our goal today was to bring you some of that the valuable insights that we're seeing within our organization as we work with different customers.

So today's agenda, I wanted to kinda start at basics and define AI in health care.

Then I wanted to talk about how AI will transform coder careers.

There's a lot of different, transformations that we are beginning to see and can anticipate that I wanted to share.

I'm just gonna run through some high level documentation and coding challenges. I'll I'll run through those slides very quickly.

If you're interested in getting a copy of the deck, it'll be available a day or two after. So, don't worry if you didn't see all the content on those slides. And then most of, today's webinar, I want to spend on talking about the potential return on investment when we implement AI and then what the eight steps are for a successful implementation.

Okay. With that, let's get started. So artificial intelligence or AI is really a machine that's helping us do a job. To me, that's what that means. And so we haven't really called a lot of the technology and tools that we've used over the years AI, but in essence, that that's what they are.

In more recent years, we're seeing AI make ways across the world, particularly with health care.

I remember, Ian, at a, old health comment, I believe, was around two thousand fourteen. So about ten years ago, there was a keynote speaker that was talking about the evolution of technology in health care, and he specifically was talking about wearable devices. So I don't know if those of you been members and and remember this particular, keynote speaker, but I was pretty wowed.

He talked about how these emerging emerging technologies will allow us to have these wearable devices on our persons, and I just thought that would be so amazing. And then here we are, fast forward ten years, and we see that. We see those wearable devices. They're everywhere. Nearly every one of us have one or or know someone that has one. So, it's here, and it's it's technology is moving very quickly.

We can expect big changes, to continue with health care and the delivery of health care. AI will reshape our daily operations, not just for the providers, and health care workers that are delivering that care, but also we'll see that within the red cycle, and we'll absolutely see it within our realm of coding and auditing.

And because of that, I think we need to embrace these changes. We shouldn't be fearful of AI taking over our jobs, and and we're gonna kinda go over that a little bit.

So I I think we should embrace it. I've been in the medical field my entire career, and so evolution of technology within health care is no stranger to me.

I find that technology is amazing only when it's developed correctly, it's implemented correctly, and it's used correctly. When those three things don't align, technology will make us work harder rather than helping us to work smarter.

Alright. So let us talk about embracing this AI technology. AI is meant to enhance the capabilities of a human coder. Right? It's not meant to replace them.

AI can handle if it if implemented and used correctly, it can take over those routine tasks and allow coders to focus on more complex and highly evolved paths? How many of us wish we had more time to do provider education or, spend time, enhancing our personal skills so we can enhance our career? I mean, all of those things can happen when we start to implement AI and and let the let AI handle the the easy things.

On the flip side in in health care, I bet you if you talk with any physician out there, any one of your other health care providers, I am one hundred percent confident that they would welcome a tool that will lessen their everyday tour of documenting into the medical record.

Right?

Lessening that burden will allow them more face to face time with patients. And let's be real. Like, that's one of the main reasons why they got into medicine is to provide that that care.

And as as a consumer of health care, how many of us would rather have the physician look at us face to face as we're having a a meaningful conversation rather than looking at the their computer and entering information?

When we're working more efficiently, our accuracy is automatically gonna increase. Our errors are gonna decline, and we're gonna, perform at a more optimal level. So think of this past Olympics when we all watched, excuse me, those Olympic runners.

Imagine if that one of the runners that was in the lead, if they were distracted and all of a sudden took their eyes off of that goal and looked at somebody in the audience or looked at somebody else, one of their components, or opponents rather, they're gonna get distracted and they're not gonna be working at their full efficiency and potentially could have lost the race. Right? Things happen when we don't have our full attention. And so, likewise, let AI do those those simple straightforward things so we can have hyperfocus on what's really important on the task that that we want to accomplish.

And, I think when we make these changes and we embrace AI, there's numerous benefits including, lowering health care costs.

So I know you've heard that before. We've heard it for years and years and years. This new change, this new shiny object is gonna lower health care costs, and, you know, we're always a little skeptical. But this time, I'm more excited about it, and I think it's more real.

I came across an article that actually referenced a paper that was put out by the National Bureau of Economic and Research. Sorry. I I pronounced that wrong. But, anyway, and they estimate that the broader adoption of AI within the US, and this is just around the adoption of health care, that there could it could lead to savings of five to ten percent.

And if you equate that to dollars, that's roughly two hundred mill billion or three hundred and sixty billion dollars annually. Now I'm always a a a skeptic, with anything that's new. Right? It has to be proven and validated.

But what if we lowered that expectation a little bit? Even a hundred thousand or a hundred billion for that matter has true potential. Right? That could really make a big difference within our organizations, especially when we're dealing with economy, challenges and budget constraints as as we have been all year.

Okay. So let's talk about the different types of AI and technology that's been out there. On the left side are some more historical tools that we've used, and on the right side are more evolving technologies that that we're rapidly seeing in the industry.

So we know a c OCR rather has been around for a very long time. We also know that, systems that can't have the ability to chronologically list things and index them in order has been around. So with the implementation of EHR systems, we've really seen these technologies kind of come to the forefront and and do those types of things, and and they're great. Right? They're very helpful. We have all also seen the evolution of voice recognition. So I remember twenty years ago working in a multispecialty practice with one of our OB docs who was brave enough to implement voice recognition, and they he used Dragon.

It was very cool and exciting, but I tell you what, as a coder trying to work through his notes, it was tough. Right? The technology wasn't quite then, quite ready for prime time and wasn't you know, there were a lot of errors and gaps and problems. And if if he didn't go back and make corrections, we were left with, I don't know what he's trying to tell me here, types of things. And so, lots of technologies have have come into play since then to make that better. For example, natural natural language processing LMPs.

When those kind of came about and, coupled that with maybe machine learning, helping these systems to learn and improve, now all of a sudden, we we're seeing some improvements, but there were still challenges and issues. There's still a lot of faults, positives, a lot of errors that coders had to go in and make sure that we're correcting and mindful of. Whereas of late, where we see this evolving AI, particularly around clinical documentation improvement tools, where they're using these tools are using a combination of of technologies that have come beforehand, where they're using these indexing and natural language processors and machines that are learning and growing over time, and they're just they're getting better, but it it's still a little slow. Right?

We see virtual scribes popping up everywhere where, again, they're taking a combination of these different, tools and technologies that are out there and combining with it human scribes and what they would do and coming out with some pretty cool stuff. Right? You've got a little AI buddy on your shoulder that's helping to describe these notes.

So imagine if you combine that with AI AI powered chart summarizations where you can take that scribe and then develop some thoughtful, meaningful summary. So, for example, we all use Teams or most of us use Teams. And when we record a meeting, it will summarize that meeting and give us action items.

We use Gong in our Salesforce as part of our sales and discovery calls with clients, and it's really awesome to have that recording and go back to, that summary of action items because we're we're only human. Right? We can't remember and write everything down. Right? These tools are there to help make our jobs easier, and they are. They're making a difference.

So generative AI is really taking, all of these different components of technology and really creating a human like clinical note and really taking it to that next level. And here is where I think the game changer is. So we're gonna dive a little bit deeper into that.

Alright. So here's a quote by Ryan Cameron, who is a president of technology and innovations at the Children's Hospital and Medical Center in Nebraska. I really liked what he had to say about AI. It's not magic.

Like everything else, we have to test it, run it through clinical trials to ensure the solutions actually work in the real world. The big challenge with AI is ensuring there is a health care specific code of ethics and a regulatory environment.

We need to ensure the use of AI is always safe and clinically proven.

So as HIM directors, coding managers, and even as coders and auditors, we cannot provide this, the same logic, right, to what we do. We're not providing patient care, but we are part of the patient care delivery system.

K? And so if we say, okay. Well, we need to test the AI and validate it and ensure that it is accurate as as far as clinical documentation is accurate for coding, based off the suggestions, then all of a sudden, it we can use this tool to help our revenue integrity. So my bot motto when it comes to AI technology is validate and trust.

Okay. Let's do a poll question, Cordell.

You got it. I'll launch that now if you wanna read it out to the audience.

Of course. Is your organization currently using AI for medical record documentation and coding?

So we wanna know, are you using it for documentation only? Are you using AI for coding only? For both? Or are you not using AI at all?

We're gonna wait till we get about sixty percent participation just so we have decent sample sets, and we're just about there. I'll share the results in one moment.

There you have it.

Awesome. So we've got fifty eight percent of our attendees that are not using AI for documentation or coding.

We have fourteen percent that are using both.

So small percentage, but it's out there. There are early adopters, and there's nineteen percent that are using for documentation only. So, I'm actually pleased with these numbers. I I love to see the evolution of technology and understand that organizations are are early adopters. Thank you for participating. So let's jump over, to the next section.

So transforming coder careers. Health care is forever evolving.

As a coder, we have to adopt to never ending changes. Right? We survived EMR implementations and reimplementations.

How many of us have gone through that? We have survived the on and then off again and then on again, ICD ten transition from I nine to I ten.

We've embraced and survived the twenty twenty one and twenty three e and m guideline changes, and we are better for it. We are better coders. We are better managers, and we are better directors. Because of all of these major evolution and changes, we are better.

I, for one, love the new E and M guideline changes. I find it way more, challenging and, fulfilling and exciting when I get down the weeds and I actually audit or code using the newer guidelines because I don't feel like a robot anymore. I don't feel like I'm just looking for words and I'm checking boxes. I actually feel like I'm part of the health care, delivery system of the of providing patient care. I feel part of it. I be I get to look at the note and paint a picture of what happened in that visit.

Where before it was it wasn't like that for me.

And so I think that the AI changes that are coming as managers and directors, we need to encourage our coders to embrace this transformative opportunity.

Coders possess skills and abilities that are exciting and perfect for the future of AR of AI rather. Coders have the strong ability to adapt. They have a passion to change and implement those changes. Like, we do it every year with the guideline changes. Right? They have the ability to be very thorough and very precise in the details of what we do.

We have to apply critical thinking even more so with the guideline changes I just mentioned, and we have to be very decisive about those decisions that we make. And so as directors and managers, let's really, hone in on these skills that our coders have and give them more challenging opportunities, more tasks so that they can advance their careers.

With this evolution of technology, coders can then advance their careers to more highly specialized roles. Right? You can, get specialized certifications, that bring about opportunities to review more complex cases. So AI, there it's gonna get really good at coding the routine stuff.

Right? But that's gonna leave these more complex cases for a coder to look at. And, it's gonna require a human judgment call, right? It's going to evolve coders into thinking more like an auditor and reviewing these cases and digging deep instead of just running through production coding.

It's going to allow coders to evolve into positions and even auditors into more, CDI type specialist positions.

We're gonna talk about this more in just a minute.

And it's gonna allow our coders and auditors to become those truly expert and, go to educators.

Right? They know the rules, they look at our clinical notes all the time and they know where those gaps are. They are the go go to people, to help us overcome some challenges that we continually face.

Okay. So complex specialties and cases.

I like I mentioned, words can only shape and outline a picture. It's not gonna always provide the color that you need to ensure the accuracy of coding for reimbursement.

So when you have specific nuances that are patient specific or guidelines and rule specific or, gaps in the the clinical record, only a expert coder or auditor is going to add that extra color that's needed to paint that correct picture.

For example, if, not always all the notes do they call out whether a condition is acute or chronic, or is it active or inactive? I mean, those are some of the challenges we face today, especially around, you know, is is the patient has cancer or is it actively being treated or not? Those are often questions we ask ourselves all the time. And so AI isn't always going to be able to handle these contextual judgment calls. Only a highly skilled coder auditor is going to be able to. And then what I love is really making our coding and auditing patient centric.

Right? Getting back to really understand the patient and the visit, what happened between them and their provider, and creating a colorful picture.

In addition to these complex cases, coders, can really bring about interdisciplinary communications.

Coders are the ones that typically work with the health care provider. They work with the billing department.

Sometimes they're asked to, you know, work with insurance companies and work through these denials. They are the ones that then can take these stakeholders, get them together, and really put together some, root causes of challenges and and problems and then use AI to help resolve for those challenges.

They're the ones that are on the front lines of knowing what the industry changes are, and they can help disseminate those throughout your workflows, throughout your EMR system, and then even helping AI companies improve their tools and systems to with what these guideline changes are.

Okay. Turning auditor or coders rather into auditors. Right? Coding is never gonna be a hundred percent correct and error free. That that's just not the way world works.

And so auditing is such a critical part of your compliance program. We all know that, but here's where we can start to transition our coders to higher level careers, right?

Auditing and validating these AI companies that are coming through is just absolutely critical to the their success and critical to implementing these AI tools with your within your organization. We're gonna talk about that in just a minute. And then the ongoing CQI process is also very critical.

Auditors just have that natural ability to look at things in a little bit different perspective and put those root causes and solutions together.

Part of that root cause and solution is becoming a CDI specialist. You don't have to be a clinical person. You don't have to be a nurse. You don't have to be a provider to be a CDI expert. Right? Experienced auditors have keen insights to documentation gaps. They understand basic disease processes and can help identify really these these, areas of opportunity and really come in and and bring that interdisciplinary team together to really make a difference, going forward and improving, the the solutions of of the AI and being kind that that liaison person.

Okay.

Auditors, because of of their skill set and experience, I I know that we've got all of us have one or two coders or one or two auditors out of our whole group that they're the go to person. That's who we always go to when we have a question. Right? Some of your physicians have have that go to person, constantly that they're always there, and they always, provide us with help and support and the answers that we need. Well, enhance and, promote those folks, right, into specialized educators.

They can help come in and really build out more comprehensive training. So if they're not worried about day to day production, they can come in and take, when we've identified those root causes and potential solutions, and they can really help implement them across your organization.

Alright. So let's talk about very quickly some documentation and coding challenges.

In twenty three, we did a case study. We looked across thousands and thousands of different, records that we audited, and we looked at professional cases, all specialties. We looked at facility cases.

And overall, we found that there were issues with accuracies still. Right? We're seeing, some unique opportunities and some new things that we didn't see. So I wanted to give you just, some interesting insights here. So we're still seeing areas of overcoating.

We're still seeing areas of undercoating and in some specialties that's higher than the average of five percent.

What was surprising to me is over the last several years, maybe three or four years, I'm seeing a steady decline in ICD ten accuracy and even more alarming, we're seeing an increase in errors around ancillary services and even missed, codes or missed revenue opportunities.

So lots of areas of improvement that we need to shore up. When we did root cause analysis and we spoke with our customers about these errors that we were seeing, we identified that a lot of it came down to inadequate or inconsistent documentation, challenges within EMR systems and templates and users. We know that there's time constraints. How many of us feel burned out? How many providers do we see that are burned out in their staff?

And that just creates a domino effect of errors.

One thing that was a little new that I hadn't even seen, even when I did a case study with our twenty twenty one E and M guidelines is an increase in missed opportunities for querying the providers. Forty nine percent increase is what we saw in twenty three. That's huge.

People are too busy, right, to take time to, communicate and query the providers, or maybe they're not getting the response that they need, and so why bother?

We're seeing a increased steady increase in errors of undercoding risk adjustment where, hey. If we're not getting getting, things coded correctly upfront or documentation isn't right, we're seeing these errors creep up.

There was a study, just last year that a f HFMA, let me get my acronyms correct, said that they're seeing a forty two percent error rate in claims, denials due specifically to errors related to time constraints, lack of automation, and inefficient workflows. I've got a link to this, study and these results in the reference slide so you can look that up. That's pretty telling, and that goes along with the errors that we're seeing.

Right? So besides documentation and coding challenges, there have been challenges with historical technology. Some of those technologies, that I mentioned before, particularly around computer assisted coding, natural language processing systems.

These systems were good and only got us to a certain point, but it they created a lot of extra work to ensure the accuracy. Right? We know CMS did a study on, these tools around risk adjustment several years ago, and we know that a lot of these systems, you can set a threshold bar of accuracy versus error, and you can play with that threshold. All those thing types of things lead to extra work, false positives, and, overall, you know, lack of of oversight, potentially, that could result to a lot of these errors. Maybe that's why why we're seeing an increase in these errors.

So at AAPC this past year, I would probably say just over maybe about thirteen, fourteen months.

We've actually been working with several different AI tech companies and validating the accuracy, the coding and documentation accuracy that is, of their systems.

And preliminary reviews of what we've looked at are pretty amazing. There's some high level accuracy rates out there.

And I I really feel that to unleash the full potential of AI in this area, it are the companies that have an AI scribe tool married with CDI and coding awareness.

Groups that are just solving for AI scribing is is gonna reverse the timeline. So if we look at this screen right here where we have just an average physician doing their job, doing their documentation, we have to incorporate change management. That's the coder coming in, shadow coding, doing queries.

We have the CDI team looking at, clinical documentation gaps, and that gives us to a more ideal level of accuracy, for coding and documentation.

If we have a a naive AI scribe tool that's just scribing, that takes us backwards. Right? So imagine if we had an AI scribe tool that was married with CDI and coding aware ness, we're gonna get back to our ideal.

Alright. So let's switch gears and talk about, some potential AI, or potential ROI when we implement AI. Okay? So back to this twenty twenty three case study that we did, we we published a lot of that data, and there's a lot of different opportunities for improvement.

So we took those challenges and some of those real were real world scenarios, and we worked with a couple of different AI scribing, vendors that actually have these the CDI and coding awareness capabilities, and we push through what the ROI potential would be using these systems. Now I'm gonna throw out a disclaimer out there. While these examples are real world scenarios, the results and the actual return on investment potential could vary. It's going to vary by organization.

It's gonna be vary by by different AI vendors. Right? But the preliminary data is so exciting, and I think that's, once you see this, you'll get gain the excitement that I have and see where there really is health care cost saving potential here.

Okay. So when we implement and use AI Scribe tools for medical record documentation and coding, there's a lot of different cost saving benefits that we can gain gain. First and foremost, the number one benefit is reduced documentation time. So I don't know that there's any physician that's gonna shy away from something that could help them.

We mentioned that earlier. But that's not to say we're gonna tell those those physicians, okay. Now that you're saving time, you need to see two, three, five more patients. We don't wanna do that.

Right? So, really, as managers and directors stress the the point with your your suite seat folks that we need to help the providers decrease their burnout, decrease that documentation burden, and increase the quality time that they're spending face to face with patients. Because when we do that, there's potential in that now all of a sudden they've got time to perform care management. They can build these relationships and potentially build for that g twenty two eleven code, that we're all struggling with.

You know, how how do we correctly document those, those services.

Now, all of a sudden we can focus on the CCI risk adjustment gaps with conditions.

Our providers can make sure they're including those with this, their their patients, that they have time to do that. So when we increase our accuracy of clinical documentation, we're gonna increase our coding accuracy, and we're gonna increase the overall compliance and reduce the risk. Right? We're gonna reduce patient safety issues. We're gonna reduce errors, that maybe the, clinical staff and the operation of your practice are going through. So just imagine if your physician's happy, they're not constantly running behind behind, they're not running a rat race.

Your staff is gonna be happier, right? And they're not gonna be as likely to make errors, miss opportunities, with, with rescheduling patients or whatever whatever the case might be with that doing those things that really make a difference in that, matter within your individual practice. And all of those things are ultimately going to impact your revenue, impact the revenue cycle. We could even potentially see, fewer claim edits and denials because if all of that's more accurate, it's likely that that that's gonna we're gonna see those results as well.

Oops. I'm heading Going too fast. Alright. So in in a white paper that we released, we really targeted, those challenges that we briefly talked about, and then we paired that with this AI scribing tool that had a coding buddy of CDI and coding awareness.

And here's, the potential of ROI for an ED visit on medical doc, MDM, medical decision making improvement.

Okay. So here's just a typical assessment plan that we might see from e r EHR templates, or we might see from a coding AI Stripe tool that's that I'm sorry.

Naive coding tool that's an AI tool that's just scribing, right, and not really coding aware. And so we see there's really nothing wrong with this assessment plan. We can, clearly see the patient presented with chest pain. We can clearly see there was workup and diagnostic tests, done and patient was admitted.

But what we don't see are the details. We don't see the color. Right? We don't know if this ED provider actually did the formal interpretation of that X-ray.

We don't know if they did the interpretation of the EKG.

We don't know what level the patient was admitted. We can certainly make assumptions, but we don't know. It doesn't tell us. We don't know if this ED provider talked with the patient's primary care provider or called in a cardiologist.

Those are all things that are going to impact, the color for that painting that picture as well as the coding and ultimately the medical decision making level. So in general, this probably is more more likely a moderate level.

Whereas, if we bring in a coding aware and CDI aware scribe buddy, we could potentially put together a more meaningful assessment and plan that would look like this. And so it's pulling the different components, summarizing what happened in the visit and putting together a meaningful assessment and plan where now all of a sudden we can pull out those components to really level this correctly. So from going from moderate to potentially a high level, we know the CMS be scheduled today for twenty twenty four, a moderate level ED visit has a reimbursement of hundred nineteen dollars, whereas a high level E and M is a hundred and seventy three.

There's a there's a difference. It can be impactful.

Alright. Let's go over an example of ICD ten documentation, specifically around HCC coding.

So in this example, this is a common statement that we might see, with a patient with chronic kidney disease where a patient presents with worsening edema and fatigue. They re the provider reviewed their medications, telling the patient to continue, recommendation for diet, and then follow-up in a month or so to reassess the kidney function and their heart failure. Well, we don't know, one, what type of heart failure the patient has. We don't know the stage of the chronic kidney disease, and we don't know the correlation, between these two conditions.

As an auditor, we can make those assumptions, but it's not written. Right? We can't add that color because it's not documented.

Whereas if we have this AI buddy that was CDI aware and coding aware, we might get something that would look like this, where we take all of those components that happen within the visit. And now all of a sudden, we can see that color, and we have higher level potential of, coding to the highest level of specificity. So we can get an HCC code category, for two hundred twenty six for CHF and three twenty nine for stage three, chronic kidney disease. When we have those higher level HCC scores, that's gonna impact the patient's RAFT score, and it's ultimately going to increase potential payment for cost sharing for with your organization and with the health plan.

K? Can make a big impact. The white paper, we go through, another scenario around facility DRG and sepsis that you might find insightful.

Okay. So that's just documentation and coding. What if and imagine if your AI scribing tool can also provide little nudges, real time in nudges at the point of care.

That's game game changing. Right?

So let's look at at some of these potential opportunities here.

So how many times as a coder or auditor we see in the note smoking cessation was discussed with the patient, and that's all we see? I wish we could all be in a room and raise our hands and see how many of us, had have dealt with that.

We know that potentially more happened. We it suggested, particularly with this patient, with COPD, that perhaps that conversation took a little bit of time, and was, really stressed, but we don't we don't know that from the note. The other problem is we're not gonna go chase this. Right?

It's you know, after the fact, are we really gonna go ask the provider for clarifying, questions around, you know, how much time did you do? Are they really gonna go back and change their note? Like, it's just something that is just so hard to chase and sometimes just not worth it. Right?

However, if we had this AI buddy that gave us the provider these real time nudges that could say, hey. Wait a minute.

There obviously could be more happening than just a discussion.

Maybe this AI could prompt the provider with a few questions that could then be answered and, embedded into the clinical note that would support coding smoking sensation. Right?

What if some of these nudges, were around the higher level accuracy adding, smoking dependent codes on on the today's visit, addressing those problems.

All of these, you can start to see could has huge potentials, within the organization.

Okay. Let's move on and talk about the eight steps to a successful implementation of AI technology.

I had the opportunity this last September to attend the Chime conference hosted by the Cleveland Clinic. Now in this instant, Chime is not a bank.

Chime is actually a financial tech company.

And over the last couple years, they've really been fostering and nurturing and building relationships with the technology industry and health care.

And of of course, they'd be doing this, right, because that's where, in there's a huge impact in the delivery system of health care that that technology can has a a major role in. So it it makes sense that Chime would do this. And this conference was absolutely amazing.

I I got to turn sit and listen to a lot of different folks, during, you know, at at different tables, at lunch, at at at the activities. I got to participate in several of the sessions.

And it was interesting to see that at this conference, there were a lot of different AI scribe companies that were there, different vendors, and there were a lot of organizations that have actually been early adopters. They have already implemented these AI scribe tools, and they talked about their experiences, what went well, what didn't.

And there were folks that were in the middle of implementations as well. So the Cleveland Clinic, it was interesting because they took a little bit of a different approach to their implementation where, you know, others had decision made and they went big bang. Cleveland took a more thoughtful approach, and we're gonna talk about that in just a minute. But out of all of these conversations and the sessions that I attended, one thing was very, very clear and absolutely unanimous across these different groups was that everybody was there because they wanted a partnership.

Organizations weren't looking for just a tool or a vendor.

They were looking for a partner.

These AI scribe vendors weren't looking just to sell their technology. They were looking for a partner to help just expand and evolve the area of technology in health care. So it was actually pretty amazing.

Alright, Cordell. Let's do another poll real quick. Let's just take a quick pause.

Alright. That has been launched.

Alright. Is your organization currently considering or in the process of implementing AI software? Yes, no, or you don't know?

Alright. We are at that magical number of sixty percent participation, so I'm gonna end the poll and share the results.

Alright. Fifty nine percent said yes, that they are in the process of implementing AI software. That is awesome. I told you there's a ton of early adopters out there.

Twelve percent said no. Twenty nine percent said they don't know yet. If you don't know, find out. Start asking your your boss. Have your boss ask their boss. Really find out what's happening so you can really, engage yourself in into that process.

Alright. Thank you.

See if I can get my slide to flip over. Here we go.

Alright.

Step number one, you need to know where your pain points are. Right? So the very first thing that you need to do is understand where your coding, and documentation gaps are. What is your current accuracy rates?

When you do that and then you start to work through some root causes, you can, start to identify your provider workflows and what their pain points are and their issues are. So when you have the the coding and documentation piece and the real user piece, then you can start develop, okay. Here's our goals.

Our here's our challenges, and here's the goal of the the AI technology, what we want to solve for.

All of a sudden, now you have very clear objectives. You can put those together.

Step number two.

Once you've identified those challenges and the problems that you wanna solve for and what your end goal is, now you need to identify your stakeholders. So we we know that from implementing EHR symptoms, not EHR symptoms, EHR systems. We know we've gotta have stakeholders.

And I think with implementing technology, this is even more so important than it is with other technology and tools that we have done. We really need to have the users, the people that will be using the system and impacted on the frontline.

We really need to make sure we're including auditors within this process, because they're the ones that are going to point out if this technology if there's gaps in that technology, if it's really gonna solve for these problems, if it's gonna create risk, and your clinical folks are gonna be able to identify those things during the patient care part. Okay? So it's important to get all of those people in the same room and on the same page.

The other thing in vetting a vendor is that you need to find out what specialties that vendor not have the capabilities of, not just necessarily what they can do or claim they can do, but what they've implemented wide. Who's using those those specialties? If you're just doing derm, maybe that's just what important for for you. No big deal.

But if you're a big health system and you've got multiple specialties, this then can become a big deal. Find out who is actually using their system live, what specialties, find out what their accuracy rates have been, find out what interoperability, systems that they have. Is it gonna work with your EHR? Is it gonna work with your other systems that your providers are using?

Are your providers using screening tools or or other things during that patient delivery process?

Ask these questions and vet these things out front.

Also, of course, you gotta get your HIPAA compliance, folks involved to make sure that these tools meet the the current requirements.

Okay. So once you've identified an, AI tech company or a couple, you wanna pilot this. So at the CHIME conference, it was really interesting to listen to the Cleveland Clinic approach, and there's lots of articles about their approach and adoption of AI Scribe technology. So I would strongly encourage you to go. Just Google it. You'll you'll be able to see.

They took an approach where they vetted many different AI companies, and they made a selection of a few handful, and then they identified some key champions. They took providers from key specialties, not every specialty. I believe it was around ten or so, and they identified champions within those specialty, and then they defined their pilot.

So remember, when you're doing a pilot, you're not all in yet. Right? You need to set up guardrails for this pilot because you don't know exactly what's gonna come up. Right? That's why we do a pilot. But if you're all in, you're gonna get down in the weeds. It's gonna be a horrible experience.

You need to scale that back just a little bit, set up those guardrails, and you need to gather data, gather data, gather data, set specific KPIs, key performance indicators and thresholds of where you expect this technology to be, set those KPIs for the individual users, set KPIs for patients who says a patient can't be part of this, this pilot, right? They love it. They, they know tech is going on all around them. Let them know what's happening, survey them, have your auditors come in and audit and re audit and audit and re audit. All of, all of those things can happen over the course of your pilot. Once you gather all of that data, then you can come up and review your assessment and what risks may come up from from your pilot, what solutions are solved, what risks are there. That's gonna help with your final decision.

Okay. Before we continue with our eight step process, Cordell, let's run this this next poll question.

Alright. Everybody should have that on their desktops now.

What kind of AI tools is your organization using or considering to implement? CACs, Scribe, coding, denial management?

Rudel, where are we at percentage wise?

At forty five percent. See a few people suggesting they would like to choose more than one option, which I wish could fill out.

Yep. Yep. We we should've we should've thought that, but we appreciate the feedback. Next time, we'll we'll get there.

Alright. I'm gonna go ahead and end the poll and share the results with everyone.

Alright. Thirty four percent say that they're implementing computer assisted coding, followed very close by scribe documentation, coding, and then denial management. So it was ranked order.

Very interesting. I I love that feedback.

I would love to hear how many organizations are attacking AI and and implementing multiple tools. That would be great.

Alright. Let's move on.

Step four, customizing workflows.

Once you've identified what AI company you want to implement, you need to go back and you need to shadow your providers and really come to understand what their workflows are. The goal with implementing AI technology is not to make the provider change how they see patient practice medicine.

In order to make sure we're not doing that, we have to shadow the providers and understand their normal process.

Each specialty is going to be different. Each pry provider is gonna work a little bit different. My recommendation is is for the shadowing that you pair a coder auditor with a clinical person like a CNA or an RN so that you're shadowing that full cycle of, the patient being scheduled and the the the coding of of the claim. Right? It's going to be very insightful.

And then once you have that data, you want to, you know, streamline the process of implementing this this technology so that you're eliminating those challenges and barriers and pain points that you previously identified and not impacting the workflow.

The workflow should be very logical, simple, and straightforward. So just imagine if you have this AI buddy with CDI coding awareness and the nudges, then you you you're gonna be aware of a challenge or a a gap, and you can all of a sudden start to implement these little nudges during particular times of those workflow.

Share these workflows with your AI vendor. Right? Doesn't do you any good just to keep them in house.

Step five, you've gotta train your providers.

So it was really interesting to hear some of the organizations that are early adapters of these AI Scribe tools.

One organization, they did a big bang. It was rough on the providers. It took them several weeks and, you know, over a month to to get on board. Whereas the Cleveland Clinic, described a particular vendor that they didn't have to do anything the night before, week before. They just showed up that day of the go live.

The the vendor provided a few minutes of training.

Someone was there the first, several patients, and that was it. They weren't asked to do anything different. They were just instructed on how to turn on this AI buddy, what to expect, and what to do, to finalize their their note.

And it was easy. It was painless, and they loved it. So this is a great question as you're vetting an AI vendor. What is a physician training time? What's what's expected of that effort?

Step seven, address change management outcomes.

So we've all been through EHR implementations, and change man management is one of those areas that oftentimes isn't really done well.

And in this process, we have to put it forefront. We need to make sure that our teams are nimble, they're flexible, and that we're there in the trenches that we're having daily scrum meetings with these practices and providers that are in their implementation and maybe twice a day in the morning and at lunch where we're gathering that data, where we can celebrate successes, we can listen to any ongoing pain points and solve for them.

Step eight and the final step is compliance, auditing, and continual documentation improvement.

We need to make sure that we're not only auditing the early stages of the outputs of these AI scribe tools, but it's ongoing, and we're incorporating these constant regulatory changes that come up, and we're constantly communicating this information.

It's never ending.

The future of AI health care is is inevitable. Right? The impact on coders is inevitable. But we can embrace. We can get excited about career, transfer main transforming opportunities, we can get, excited about the potential of return on investment and the potential of time savings and low, decreasing burdens and pain points.

I I really believe this to be true. And, I really feel like embracing, you know, as management and directors, if we get excited and we embrace these things and and we help our staff to do likewise, it's just going to result in a more successful implementation of these tools.

Alright. I we have got one minute left, Cordell. Do we have any questions, or what do you recommend we do?

Yeah. I think we've got time for one question, last minute.

And that question is, what specific documentation is required in the medical record when a provider uses an AI scribe? Does that need to be noted? Is PT consent required?

I don't think patient consent is required, and this this was a question that was brought up at the CHIME conference, and there wasn't anybody that said there was a a regulatory guideline around that. However, we want to involve the patients. And so anytime you're using new technology, it's to the benefit of the provider to say, hey. This is what we're doing.

We're piloting this. Would you like to participate? And give them an opportunity to to say no. But and I I haven't found any regulatory requirement around that.

Okay. I appreciate that. Thank you so much, Stephanie, for the presentation.

Thanks everyone for attending today.

As we mentioned, we will be sending out a copy of the slides and a recording of this presentation in just a couple of days, so keep an eye out for that in the inbox.

Also, if you are interested in having Stephanie and her team consult, vets and help implement AI solutions, or if you're interested in learning more about AAPC's third party coding and auditing services, you can do so in the survey that will pop up in your browser as soon as we leave here. Thanks again, everyone.