8 AI Adoption Risks You Can’t Afford to Ignore

Stephani Scott: Welcome, everyone. Thank you so much for taking time out of your busy schedules to join us today. I I know how hard it is to take time to, listen to to what we have to offer. And I know you're going to really, appreciate some of these insights that we're bringing you today.

Very excited to share it with you.

I've been, in the industry for a very long time and around different technologies from the beginning of EHR systems, and most recently working with quite a few different AI vendors over the last couple years. And so I've been gathering some of that information and intel.

I've also been, trying my best to keep on top of what's been happening in the industry and, attending different webinar sessions like this from different organizations and vendors and reading up on, you know, what what's been happening. And I'm out of all of that experience and knowledge, I'm I've been gathering this information and going to share that with you.

So what I thought we could start out today is talk about the evolution of AI in health care. So we know computer learning has been around for a while, but really specifically what's been happening in health care. Then I thought we could jump into why AI hasn't been as successful in some areas, and that will lead into the eight common risks that you can avoid. And I'm also going to talk about not just those risks, but potential, ways that you can mitigate those risks.

Alright. So let's jump in.

I've always been involved in health technology, tech or health care technology, for a very long time as I just mentioned. And I've always believed that technology is amazing as long as it is developed, implemented, and utilized correctly.

Technology is a means to help us, not something we can just implement and forget about.

So let's first begin by reviewing the evolution of artificial intelligence, which really is another way to say technology that enables computers to perform advanced functions and tasks.

These tasks that we're we're going to be focusing on today will include translating spoken language, analyzing data, and making recommendations.

Now the recommendations of AI technology around documentation and coding are not absolute. It it's they are recommendations just as the word implies, and so we need to keep that in mind with all of these different advances that are taking place. Okay. So let's look at this timeline.

I put this together and, it doesn't include every event, so we're going to leapfrog and and fast forward a little bit. But these are kind of the pivotal points in history that I I thought were meaningful and where there were kind of major shifts.

So if we look back at nine in the nineteen fifties, this is where researchers started to begin building computer algorithms or AKA cybernetics, like, we heard we hear that word a lot. And they started to build these algorithms to diagnose medical conditions. So way back in the fifties, that was kind of the birth of health care AI, and we know that they they really weren't all that successful back then. They just didn't have the advance to, advances in technology that we do today, but they didn't give up on on the this concept.

I didn't, by the way.

Data scientists were moving forward and and researching, and it wasn't until the nineteen eighties and nineteen nineties where we started to see these expert computer systems coming out. And a good example of something like that is natural language processing, the NLP's.

And so they they started to come out and do be, provide little helps and they were getting better as they were learning and but they still weren't perfect. They weren't great. Right?

Then let's fast forward a little bit to twenty ten.

This is when technology advances really allowed for new computing powers, and these powers were faster, more reliable, and they were different than the expert system platforms back in the eighties and nineties.

Then what happened in twenty fifteen is as these computer new computing power systems were getting a little better and wrapping up, ICD ten implementation hit. Right?

And as we started to see our documentation, our clinical documentation, and our coding, as an industry, we started to see these clinical gaps. Right? ICD ten has a much higher level specificity than ICD nine did, and so our documentation all of a sudden has the had these gaps.

And these computer models were recognizing, hey. Wait a minute. There's we we need a better way to, one, identify those gaps and then mitigate for those gaps. And so if we fast forward a little bit more into twenty twenty, now all of a sudden, we see the technology where different systems are coming together. Natural language processes are are much better. Computer assisted codings are better. Voice recognition was finally getting better, and we started to see see that coming out in different AI scribing tools.

And these these different platforms and and systems and advances were being combined and creating large language models. And so you'll hear that that term LLMs quite often.

And that's really became the the birth of Ambient Scribing AI.

And then a a little jump into twenty, twenty twenty two in late of, that year, I believe it was November, OpenAI or chat GPT hit the scene. Right? And now all of a sudden, Ambient AI that is doing scribing and doing very good with, clinical documentation, we're able to add this other layer component to enhance the computer assisted coding to more of an autonomous coding platform. Right? And so you can see that slow progress in the beginning years, and now we're on this fast track rapid progress towards this technology.

With all of these different technology companies that are out there today, there are a lot of claims of huge cost savings around AI technology, savings around claims, savings around efficiency gains and things like that. And there's a lot of different numbers. Like, you could look at one survey, for example, from MGMA that they released just this last November. They claimed a ninety two percent operational efficiencies are gained among organizations that are implementing some type of AI technology.

There are other claims that there's fifty percent improvement. There's claims of thirty six reduction in physician burnout with the, Ambient AI.

Thirty percent reduction in claim denials. And even with our, validation, we've been validating different technology platforms for the last couple years In a case study that we released just this last April, we announced in some instances of the Ambient AI with coding aware capabilities is we're seeing as high as ninety six percent accuracy.

Now with all of these returns on investments and claims of success, you're probably asking yourself, are these results real?

Well, I say they are, but you have to look at the context in which these case studies were performed, the organizations that did the surveys, who participated in in those surveys, and what type of technology platform they're they're using because there's never a one size fits all. I always say that health care is too complex, so you have to really understand the context of these results.

For example, you could take an organization that their, KPIs around the denials are through the roof.

Let's say it's like forty percent, and they implement some AI technology and all of a sudden they they gain that thirty percent, reduction. Well, that to me is believable. Right?

Because they had an opportunity where they could, implement some technology and change some workflows and gain some efficiencies. Whereas maybe another organization that has, industry standard, denial percentages, maybe between five and ten percent, well, maybe they're not gonna see as high as a thirty percent improvement.

Right? So you really have to look at the context of these results and understand them. But but I think they're believable. Right? We are seeing successes out there.

Okay. There are investments in artificial intelligent intelligence specific for health care, and these investments are soaring, particularly just as, past decade.

Investors have been pouring, nearly thirty billion dollars into health care AI startups just in the last three years alone and roughly about six sixty million in the last ten years.

So these stats came from a new report that was just published from a venture capital firm, Blair Capital Partners.

And, again, you have to look at the full article to understand the context of of these numbers. But in general, we know we know there's a lot of money that's being dumped into AI technology around health care.

And by the way, all of these stats and references that I'm listing is at the end of the slide deck, with reference links. So if you if you want to look at the data further, you can, once Cordell releases the the slide deck in a few days. Okay. So why why is the industry dumping billions of dollars in AI investments?

Well, why is AI needed in health care? Let's ask that basic question. To me, it's pretty simple because we need a fix.

We must reduce the increasing number of claim denials that are occurring. We must lessen the physician burden on, documentation and their overall workflows.

We must improve the accuracy of clinical documentation and efficiencies of our coding.

These are pain points that have always existed, but they're exponentially, growing. Right? We're really feeling the the pain of these items, and so we we do need a fix.

AI is already transforming health care.

There's lots of different studies and lots of predictions out there.

One prediction is in five years, AI technology will be implemented across ninety percent of health care organizations.

It could be different types of AI. It might not just be, ambient scribing AI or computer assisted autonomous coding AI. It could be different AI tools that are out there. And one of one of these areas I find super interesting is that the prediction is that, AI would be able to read scans and interpret results to detect more than a thousand different diseases.

So if you imagine that an AI, can look at an image or, read an MRI or other screening tools better than the naked eye, that kinda makes sense. Right?

Another prediction is that physicians may use clinical bots to help develop better care plans.

Patients are complex and sick, I for one, I can't remember every coding rule that's out there I need a help, why would we expect a physician to, remember every possible care plan and what might be best for a patient. Right? There's some great physicians out there, but they're human. Right? Just like we all are. We're all human, and, these advances in technology can help as long as they're developed, implemented, and utilized correctly. I believe that AI will never replace a physician just like it never will replace our coder or an auditor, but I do believe it can augment our duties.

Okay. With all of these technology advances and and investment dollars, the rush to implement AI in health care, why do we see AI implementations that aren't successful?

Well, like I kind of alluded to earlier, implementing AI in health care is hard. Right? AI has been successful in some areas of health care. We have seen that. Those results that I showed early earlier have seen that, but that's not without some failures.

Right?

AI is really good at repetitive easy tasks, but we're looking for more complex, tasks and functions, and looking for those more complex recommendations from from, AI. And there's there's reasons why there are failure that I think you'll find that makes sense. So let's kinda jump into those a little bit. So as I was leading to, we have to expect some failure with AI implementations.

There's several different case studies that have been published that are been out there there about some of these failures.

The most prevalent ones have been around, that radiology technology I just mentioned kind of in five years advances that might come come in. There's been a lot of failures around those and around some diagnostic tools, but when we fell, we learn from them. And then we do run another pilot or we do another implementation, and there might be some other failures and other lessons learned that we can do. And so it's okay that there's failures. Right?

It's kind of like a clinical trial. You can't just release a new drug.

There has to be some clinical trials that that are performed. So we can learn about those impacts and learn about things that that we didn't know before, and we go back and and we re rework things. Well, technology is the same way. If we don't try to implement them, if we say, hey, I'm just gonna wait until it's better.

Well, how how do we learn from that as well? Right? So this cycle of implementing or piloting the technology, identifying the failures, learning from those and trying again is expected and should be embraced, but not without understanding of some of the risks and challenges. So these like I mentioned, these failures are expected because health care is complex.

So Bevin Erickson is our CEO here at AAPC, and we were in some meetings earlier this week, and he said that he feels that health care is one of, if not, the most complex businesses in the world today.

And I believe that, every country struggles with how best to deliver healthcare, how best, is, you know, what rather, what is the best payment model?

We struggle with different challenges that come up. And let's face it. If we didn't have these complexities, we would all be out of a job. Right? It would be easy. Anybody can do it. Well, that's not the case.

And there's good reason why there's there's different complexities. So in, you know, my participation with these different vendors and my past experience and then, you know, learning about what's been happening in the industry, I've kind of broke it down to these main challenges or complexities, these these four main areas that I wanted to share. So our industry is complex. It's highly regulated. There's often conflicting guidelines amount among payers in different states.

Health care regulations and guidelines are changing. They're evolving over time and sometimes these technology platforms it's difficult for them to stay on top of that, it's difficult for us but yet, you know, they not only have to be aware of these changes but they have to build, build them into these platforms, and it's hard.

And, you know, if if these platforms aren't updated, it could result in errors, and these errors could result in penalties and there could be legal ramifications that could happen from that. And so there's high stakes which highlights how important it is to have human involvement and oversight at all stages of building these complex tools, implementing them, and monitoring them.

Another aspect of the complexities and challenges are developing algorithms for health care's complex.

Data scientists and developers have to work together and build these combination, platforms that can look at medicine and disease processes, the uniquenesses of each patient, uniquenesses of different specialties. I mean, there's layers and layers and layers of complexity that these models have to build. So it does take some time.

It's just not straightforward programming or algorithm rhythms. Right? It requires a human and their expert skills to be part of these formulas and these algorithms. Right? We've gotta be part of part of it.

Achieving consistent and reliable outcomes is also lacking a little bit. You know, primary care is not the same as cardiology or wound care, it's just not a one size fits all, patients aren't the same, workflows within different organizations are also very different And so not only taking those regulation and those other challenges, but making sure that, the recommendations are consistent and reliable is also challenging.

So developing these tools are it's hard. Right?

And the current models are do are getting better.

They're they're the technology advances are making it not as complex to roll these systems out faster, particularly with this open AI.

And, but there are still challenges beyond that, but it's implementing these tools within our existing system. So how many different systems as a coder or auditor are you using throughout your day? Right? You're using your encoder, you're using the different EHR, different specialties, locations. You've got to jump around your if if you're doing, any part of the the rev cycle, edit denials, things like that, you might be using something else. I mean, there's just a lot of different, other systems that have to to come into play and all work together nicely.

So it it makes for challenges.

Alright. So let's dive into these eight common, risk areas that, come come with implementation.

Number one, and I I rank this number one purposefully, and that is that whenever we're implementing technology, particularly AI integration, we can expect or or we definitely are fearful of a temporary slowdown of operations and perhaps a reduced productivity, right, during these these different phases that might happen.

Staff may resist change.

We're human.

It's hard to change, and that's okay. But sometimes we're a little fearful, because we know that there's gonna be some short term increased workloads and uncertainty.

Sometimes staff are even fearful of technology that, they they're at risk of losing their job.

So leadership has to deal with all of these challenges and realizations as well as transist all the transition logistics and the morale of all these teams and the technology putting it all together.

So these work disruptions are real. To mitigate them, look for an AI tool that it allows your users, particularly your physicians and other key folks, where they don't have to change their basic workflows of delivering health care. Right?

A a tool should be augmentative. It should be helpful. It shouldn't rock your world and turn it completely upside down.

The organization should be very thoughtful.

Just all of a sudden showing up in in a particular department like dermatology. Oh, we're gonna go live with your AI today.

Not a very good thing. Right? Be thoughtful and very organized about how you phase in your implementations and how you're preparing everybody that could be involved.

A lot of organizations might outsource the the AI management of these implementations. Some really good vendors, they have they already have these management teams in place that they can bring on. They're very familiar with the technology platforms. They've been doing it across the country, so they've they've learned, and they're very, very helpful. You could also have a third party vendor come in and help and maybe help with some specific areas around maybe clinical documentation or the accuracy of coding. Because the worst thing that you wanna do is have it be over disruptive so you can't get your day to day operations in. So having these experts come in and help for a short time, can just add to your your success.

And then we always hear about with with any change and particularly around, AI implementation, you wanna celebrate your successes daily and you wanna talk about what didn't work right maybe and and make some tweaks so that you're progressing quickly daily, not, slowing the that whole progression down.

Okay. Number two, there's a lot of lack of trust in the AI expertise and vendor integrity that's out there. It's because there's been previous attempts and, that that have failed. There's been a lack of understanding of the technology and this these new tech stacks that are coming out.

And, you know, it's kinda hard to believe some of those results. I get it, and I think these vendors get it too, but we can alleviate some of that lack of trust by doing quite a few different things, and we'll talk about that a little bit more. So these sometimes these failed implementations, maybe we've experienced with with different, EHR changes or other little, tech stacks tech stacks rather lead to some skepticisms that we might have. Right?

Sometimes these technology vendors, maybe they don't understand who we are and our health care systems and and deliverables and our rev cycle challenges.

Sometimes there's internal teams that lack understanding of really what the technology is meant to do and how it's meant to help. So the all of these things are pretty easy to mitigate through transparent communication.

In the last three years, I've been working with several different AI vendors, And one thing that I've learned is they they want to work with the with organizations, learn from organizations.

They really wanna be helpful.

And if we make sure that we're opening those lines of communication and we're doing our own due diligence, we can identify who that best vendor might be that can help with our particular pain points. We have to learn to ask the hard questions and ask them upfront. We also have to educate our internal teams.

There could be some skepticism and biases that are already in place so we need to address those, head on in order for us to move forward and make the the best choice and ensure that we're we're building that trust.

Okay. Number three, worries about implementation costs.

There are going to be upfront cost to implementing any any AI tool or technology, and that can put a put a a burden on our already strained budgets that we have. There's concerns around uncertain return on investments or, that that may result in delaying the decision to implement these these tools.

There could be fear of unexpected costs. Like, we we all know that there are going to be challenges that maybe we didn't plan for or cost that we didn't plan for.

But we've been through so many changes and so many technology changes over the years, you know, learn from those lessons, you know, monitor, ponder rather on those experiences, bring folks in that have recent experiences, and remember that we are going to have to, invest some money into our business so we can solve for those challenges that I mentioned before and, we can gain, some ROI on the back end.

But just remember that you need to focus on where your particular pain points are.

Different organizations are gonna have different budgets. There's going to be, different goals that come out.

So one one success story of an organization down the street may not be your success story because you may have a different pinpoint. I had a boss years ago that would say, Stephanie, we can do anything, we just can't do everything.

And so you need to keep that in mind. Right?

Identify your biggest pain points and focus on implementing a tool that can help augment and fix that pain point and so you can see those gains sooner than later.

Okay. There's apprehension about the technology complexities and compatibility with your existing system. This is also a real concern and something that we need to be aware of going on. So anytime that we're implementing new technology on our existing old outdated legacy systems, it's complex.

Challenging.

So just to share a little story really quick, a personal story.

My husband and I, we we're not the the type of folks that will go out and get the very latest cell phone. Right?

We always, you know, try to and hold on to it as long as we can before we do our upgrade. Well, my husband recently, wanted to download an app on his phone that showed, different trails.

And he wanted something with some new capabilities of from something he had used in the past. So he purchased this subscription and guess what? When he went to go download on phone, it wasn't compatible. Subscription, and guess what? When he went to go download on phone, it wasn't compatible.

There was no workaround. He had two choices, either cancel it or go out and buy a new phone to use this new technology. And so while that's just a simple story, we need to keep that in mind. So to mitigate for these compatibility concerns and, to mitigate, discouragement, we need to bring in our IT folks. We need to include our architect guys, our, our developers, our customer care teams as as it all relate to IT so that they're understanding what this new technology is and asking those hard questions that maybe we aren't, skilled enough to ask.

And, hopefully, if you can get an NDA in place with this tech vendor that they can help you look under the stack a little bit. Right? And look for those different income compatibilities that might exist. There's always a way to map map them. Right?

And there's decisions that might have to make, like, my husband, oh, I'm just gonna make, the decision to to purchase a new iPhone.

So make sure you get those NDAs in place and include your your IT folks. Very critical.

Another concern, and risk that might come up is around data privacy and security. So compliance is complex. Our HIPAA, SOC two's, our health trust, all of these items are highly regulated and they're nonnegotiable.

Right? We have to comply with these, requirements and we need to make sure that whatever AI technology is coming in can and won't introduce new vulnerabilities and increase our risk for breaches.

Patients might have concerns as well. So to mitigate, just bring in those privacy and security officers into the process from the very beginning.

Making sure that they're part of that vetting process and the decision process.

Ask questions like what happens with your PHI if it's being recorded? What happens with those codes? Where does it live?

Talk with patients, be transparent with pay transparent with patients. When they come in, tell them, hey, we've got this new technology. Would Would you like to participate? You don't have to have a a signed document from them. Not right now, anyway. You don't have to have an an attestation, but those things are nice to have. I think overall from what I'm seeing is patients are kind of excited about this technology.

So as long as you you're transparent with them and you're bringing them into part of this health care delivery system, I think they're gonna be supportive of it.

Alright. Number seven. Let me make sure I didn't oh, I did mix skip one. Number six. So concerns with the accuracy of the AI results and recommendations.

I I really considered moving this up in the list of risks, but I decided to keep it kind of in the middle because with all of the validation that we've been doing at AAPC Services, the technology has come so far.

And with that process of piloting these tools, expecting to see some failures, learning from them, and building upon them, I think that that this this risk is a little bit lower down the list. Right? We we absolutely need to be mindful and have it have it, an integral part of our implementation process.

But I think we can, once if we've identified a trusted vendor and we know what our goals are and the goals should be realistic, not everything, just, you know, this is what we we want to do, then I think we can, mitigate some of, the risk of accuracy. Right?

AI is meant to augment. It is not meant to completely take over everything. A human has to be involved.

So when we're implementing AI Ambient Scribe technology that's coding aware, look for vendors that allow for providers to review their note before they hit complete.

Right? There's no difference between if they're using an EHR system templates and smart macros and they have to review and do documentation by exception, you know, update and add whatever is needed.

That workflow isn't gonna change. So make sure that that technology allows for that workflow and that that workflow is easy. Right? We don't wanna do challenging or that just defeats the the purpose of lessening the burdens on on physician.

Making sure that your coding accuracy is correct.

Start small. Right? Don't expect AI to call the entire note, Right?

The entire visit.

Implement those, those, technologies advances slowly so that you're monitoring and you're mitigating for those failures if they come up. So, one thing that I've learned in the the recent years is that organizations that bring in a third party vendor to that is highly specialized in some of these areas can really help the organization not have to worry about some of these risks particularly around accuracy or misleading outputs.

They can let this third party vendor kinda be that angel on their shoulder and work with them so that it's there's it it doesn't have as high level disruption, right, as they're going through increasing the accuracy of this the tools that they're implementing. The other thing that you need to do to mitigate this risk is you need to review your state laws.

So during my recent research, I came across an article by the governance institute and they published, some interesting facts facts around specific states that are passing new laws around AI and health care.

The reason why we have to have these laws is they're, you know, it's a high cost. If we if there's an inaccurate claim, we all know what happens with that.

And there's some ethical concerns and risk around the data quality of of the clinical documentation and coding. Right? So if things are undercoated or overcoated, those are quality issues that we tend to make decisions and we're financially, reimbursed for. And so there are more and more states starting to come up with some of these regulations. So make sure that your team are going out and you're looking at, these regulations.

There are, some class action lawsuits happening right now around some of these health care, AI technology tools have been implemented in the past. And this article, referenced them and kinda talked about that. So interesting read if if you wanna go out there, and the the link to that article's, at the end of the deck. So number seven, this is, financial risks. Right? So we kind of talked a little bit about that, but I wanted to dive a little bit deeper.

High implementation and customization costs are to be expected. But if we plan ahead, we could maybe get ahead of of some of these. We just need to make sure that, we understand the risk of a possible delay in return on investment.

AI isn't going to be an immediate fix.

We need to remember that it's over time there's going to be return on investment, And we need to expect that there's ongoing maintenance. So, you know, you don't just buy the technology, implement it once, and you're good to go. This technology has to be updated periodically with all the changes in in the guidelines and the industry.

There's gonna be new advances that coming up. There's gonna be your your EMI EMR systems that may have might have new, technology advances that now has to your AI has to have those advances, so they play in the sandbox together. Right? So expect to have some financial, impact that's ongoing.

But if you plan ahead and you budget for those things, you're gonna mitigate a lot of surprises. The other thing that you have to be super mindful of is there's monetary return on investment and there's nonmonetary return on investment. So think about what your goal is. Is your goal to decrease have that thirty percent decrease in denials or is your goal, hey. We have a big problem with physician burnout and physician high turnover.

So if that's your main goal, then decreasing that physician turn burnout and turnover is kind of a nonmonetary return on investment at least initially.

But in the long run, you're you're gonna see those benefits play out. Same thing with coders and auditors, same thing with with other clinical staff. I mean, so you you have to look at just not the financial and monetary, return on investment, but the nonmonetary that overall could come into your, impact your finances.

Okay. We're we're coming to the to the end here. Number eight, vendor risk. So dependence on a single vendor may limit flexibility and your negotiating power. What do I mean by that?

Some vendors may oversell their features. Right?

That's the world we live in. I don't think vendors intentionally do that. I think that there's overzealous sales reps, and I think that vendors say, yeah, we can do that. But I think sometimes what they're really saying is we don't we may not have that capability right now, but we will in the future.

We can we can build it for you. So it's all about how you're asking these questions so you're not setting yourself up for disappointment and failures of of your expectations. You need to make sure that what you're understanding is is very clear to what these vendors understandings are. So in my experience in this area, it's all about when a vendor says this, I may misinterpret it to something else, and that's how, there's there could be a disappointment, right, that could lead to a vendor risk.

Make sure that there's open transparent communication to mitigate just, knowing that maybe there's, lack in a particular capability. So for example, maybe this vendor doesn't have AI scribing for a particular specialty.

That's not a bad thing. Right? That could potentially be a good thing because a good vendor is gonna work with you. They're really gonna become a partner and want your expertise in perhaps helping them build out that that specialty AI. Right?

And that's that's a benefit because then you all of a sudden you have negotiating power, right, and all of a sudden you're building this partnership and you're both in it for the long term and you're building trust, Right?

One of the other vendor risk is overselling features. We kinda touched on that a little bit, but, more so I wanna bring it into overselling understanding of of what their tech stack is. So a lot of vendors that I mentioned early earlier uses a lot of they use a lot of different types of technologies, and they may use a third party vendor. So you'll want to ask that question, like, what third party vendors are are you using within your platform? Right? Some of these AI vendors are using, like, IMO or other, coding flat platforms to, kind of leapfrog that AI technology in the autonomous coding. So ask, what are these third party text stacks and tools that perhaps they're using understand what they are?

That's also going to build trust with your vendor relationship as well when you understand these and you're like, oh, I am familiar with that.

Because a lot of these tools have been from been out here for a while. They're just getting better and better.

And finally, a good vendor will have an implementation plan beyond the first phase. So, ask what what their customer support is. If it's, hey, we're gonna get you through this first round of implementation and then it's all you, maybe you need something a little bit longer than that. That's an area where oftentimes we see some failures with implementations, because we think, hey. This initial little pilot went really well. We can do it on our own, and that support stopped. So make sure that, you're planning ahead, extending that implementation if you feel like you need that, and then maybe, pulling in maybe a third party expert to kinda help with that a little bit.

I find that when we're we ask a lot of questions upfront and we go into these situations wide open, yes, we're gonna have some unforeseen challenges and risks that might come up, But if we understand what they are upfront and we're asking those right questions, it's really going to mitigate a lot of these these risks and challenges that come up. Oh, here I go again. Okay.

I am sure we that you all agree that in health care, we need better technology to keep up with all of these industry demands that we face.

The demands of the aging population, increasing denials, increase in the OIG monitoring, they just came out with their, risk adjustment monitoring plan, a few weeks ago. All the regulatory changes that come up and our economic changes that we're all facing, not only in the health care industry, but on a personal level, these are real challenges, and these are, a perfect opportunity where we can look to technology to help us. It is clear that AI will be an important factor of the future of health care. It's not gonna go away.

We can't, put our head in the sand and let others move forward where we're left behind. We need to to at least, be monitoring and thoughtful and begin our planning stages even if it's not immediate, but we need to be moving forward, taking some step forward. AI isn't just about technology and doing work for you. It's about helping you do the best job that you can.

Successful AI adoption requires thorough validation, continuous monitoring, and expert oversight by medical auditors, CDI teams, and all those other experts that I that I spoke about.

Implementing third party vendors to come in and help you with this is a smart move and, can really lend to the success so that you're mitigating some of these risks that a lot of the other other organizations have gone through.

Alright. Well, I think we have a few minutes, Cordell. Do we have any questions?

We sure do. Let's jump right in. We've got a few minutes to cover them. Great. The first is, do you are there any apps that you would recommend for researching health care questions, particularly particularly for a small multidisciplinary medicine clinic?

Is there is there any that you could recommend? Whether that's app or or software, I think.

Yeah. That's such a a great question, and I think this question came up the last webinar we did in April when I talked about our twenty twenty four case study results, and I included some of the AI case studies in there.

You got you they need to be trusted sources. So a OpenAI is out there and it's everywhere.

And as long as you're asking the question in the right way, there are good recommendations.

So I I don't know that there's one particular app that I would say over another. I think there's, over time as you're using OpenAI, it gets used to you and your questions and and things like that. So, we're we're still at the stage where these recommendations are just that recommendations.

What we you tell our team is you still have to go back to the source of truth, go back to to the guidelines, go back to the code set rules, and validate what those are. So even though maybe it appears that some of these apps are providing you with a great recommendation and they even tell you, oh, here's a reference. You still have to do your diligence and ensure that that it's correct.

So if you are using a particular app, get to understand, the tech stack behind that and how that vendor is is you know, what tools they're using to substantiate those those recommendations. So I know it's not a a great answer, but, you know, there's there's some good tools out there. You just have to make sure you do your due diligence.

Alright. Next one, and this is a great one. How should auditors and coders prepare to arm themselves to be integral to the AI implementation process? What kind of skills should we gather or acquire or courses to take?

Oh my gosh. I love this question.

I I think we need to embrace the technology. Like I said, it's it's not gonna go away.

It's already here. And I think the more that we can educate ourselves as coders and auditors, look to our our management team and and say, hey. Look. I I I we need this education.

We we need to stay on top of what's happening in the industry because knowledge is power. Right?

So the more that that we can get that education and arm ourselves, the better. I think that we need to, as auditors, be mindful of where our gaps are. We are not perfect, in our skill set. Our workflows may not be perfect. So kind of by identifying where those gaps are, as an organization and as a team, and then understanding how technology perhaps could help with those gaps or perhaps where it it may not be necessary. So I just really think that arming yourself with a much as much information as you can and then sharing that with your boss, with your physicians, with your IT team, making sure upper management, your leadership really, understands that you're staying abreast of all these things as equally important because they're going to, gain trust in your skill set and understanding what's happening and bring you into those conversations of vetting and trust what you're saying and telling. So all about understanding, gaining knowledge, and transparent communication and trust.

Cridel, do we have other questions?

Sorry. I was on mute. I was just gonna add because you've said it in the past is to be a first adopter of any software solution that your organization implements. Be the power user, the subject matter expert, etcetera, so you can teach others.

Absolutely.

Next next question. How can I effectively communicate to my manager that a particular solution may actually slow down our production as opposed to improve production?

You know, that's another good question. I love you guys. This is great.

You gotta have your your your facts. Right? Whenever I bring a concern to my boss, I've gotta have my ducks in a row. I gotta have facts. So if it's just a concern, voice that concern and then figure out a way together to, to either validate that concern or devalidate it. Right?

Go through that. So either maybe get some references of from a certain vendor, look for case studies, you know, doing pilots is a good way to say, hey. I'm I'm a little skeptical. I'm concerned about this this technology.

Could we perhaps just first do do a pilot? So, you know, and and maybe in the past that particular technology hasn't been great and maybe they've make it may have made advances and it's better.

So it's it's getting your hands in and really proving is it is it gonna work for us or not is the is the best way to go.

You could also share with them, your manager this presentation.

Yes. You could.

One final question I think we've got time for.

If we started using Coding AI software, how does that change our role as a coder or auditor? Are we just verifying what AI wants to submit on the claim?

It depends on what type of AI you're using. You know, old, natural language processing, you're you're agreeing or disagreeing with what it's presenting with you or adding more. I think the advances nowadays are more than that. For example, the different organizations that I've been working with that have implemented ambient AI and scribing, with the the computer assisted coding or autonomous coding.

We're we're reviewing that clinical documentation for accuracy as well as the coding, and that role hasn't changed for us. As auditors, it hasn't changed. We're doing the exact same thing we've always done. We just have a different understanding of what we're looking for, how that documentation is created.

So remember, if you join me in other webinars, I talk about documentation coding. It's all these pieces from the the visit that we have to put together. We were present in that visit, so we've got these different pieces to the puzzle we have to put together and and come up with the best outcome that we can for coding it. So as auditors, it's the same way, but we've got other pieces here in the technology.

So I have found on a personal level, what I do hasn't changed. The work hasn't changed, but my perspective and understanding and how I'm communicating the results to the provider is different, but the work itself isn't. And so I would imagine at a as a coder, it would be similar. I mean, we we can let go once we build that trust and validate, let go the easy stuff and focus more on the harder.

So we're still doing the same process.

It's just going to be either the the harder more challenging or the messaging and things that we're looking for is just going to be a little different.

Alright. Thank you, Stephanie, and thanks everyone for joining today's webinar.

If you are interested in learning how Stephanie and her team can help, audit or help implement an AI solution, you you'll be able to let us know as soon as you leave this webinar. You can also explore other revenue cycle solutions like her third party audit and coding services to add that human element to your revenue cycle.

Thanks again, and remember, we will be sending a copy of the slides and a recording of this presentation out early next week, so keep an eye out, for those in your inbox. Until next time. Thanks, everyone.