Key takeaways from this episode

• CMS has directed the MACs to hold Medicare claims for 10 days during the shutdown so they can reprogram their systems and process claims once. Keep submitting claims; the MACs are queuing them, and Medicare itself is funded and not part of the budget fight, so the delay is administrative, not a funding problem.
• Until telehealth coverage is resolved, get an ABN signed and/or reschedule Medicare telehealth visits. There are rumors that telehealth may be reinstated retroactively, but PRS has not been able to confirm them, so the ABN-or-reschedule recommendation stands.
• Treat AI as a protocol that must be tuned to your practice, with human supervision from people who have the subject-matter expertise. The computer learns in a linear fashion, so an early error can derail the process downstream; pick partners with coding and practice expertise, not just programming expertise.
• A vendor's "99% correct claim rate" measures clean claims that go through, not claims coded correctly for what was actually done. Historical paid claims may have been coded wrong to begin with, so ask the right questions and dig into any savings numbers before you buy.
• Ambient listening tools only hear what is said, so visits documented with them can drop from level 4 to level 2. The tool does not see orders you click or data you review; understand what the tool has access to and interact with it accordingly.
• Even with "AI in the loop" handing tasks from one AI to another, a knowledgeable brain is still required. Documentation, interpretation and coding are siloed steps, and a person who understands the system has to check the results at each stage.
• Experiment with AI, but monitor it, and expect to re-tune it for every practice. A system that works for one practice has solved the problem for one practice; staff workflows, system interfaces and delegation differ everywhere.

Transcript

Lightly edited for readability; timestamps mark where each speaker turn begins in the recording. UCR 261, recorded October 2025, with Scott Painter, Mark Painter and Dr. Ray Painter of PRS Network.

Government shutdown, day three: the CMS work plan and the 10-day claims hold

[00:00] Scott Painter: On this episode: the latest on the government shutdown and how it impacts urology practices, and a human perspective on AI. Stay tuned. Welcome to episode 261 of the Urology Coding and Reimbursement Podcast. I'm your host, Scott Painter, with my co-host Mark Painter and Dr. Ray Painter. We wanted to first cover the government shutdown, what it means, where we are, and what to do, or at least our recommendations on what you should do. Let's cover that quickly, and then we want to talk about AI and our observations on AI as Mark and I have been traveling around the country. So first, the government shutdown. Where are we, Mark, and what does it mean?

[01:00] Mark Painter: We're on day three of the government shutdown as we're recording this. You're probably all aware of the positioning that's being taken in Washington relative to this, and the finger pointing and the blame game that's going on. Unfortunately, what we're not seeing at the moment is progress toward resolution. I certainly hope somebody can go above the fray and get this thing over with. In addition to healthcare and everything that's going on, this is meeting season and travel season, and of course TSA and the FAA, all those folks, are currently working without getting paid. They are expecting to get paid at some point in time, but they're not going to see their next paycheck, which is not comforting.

From a healthcare perspective, we did see, and I think we put a notification out, that CMS put a work plan together. They dropped it, of course, after the shutdown occurred. The one thing that we missed a little bit from our perspective was that CMS directed the MACs to put a hold for 10 days so they could adjust their systems and not have to double-process claims.

I've heard rumors in the background that there are some moves afoot to potentially reinstate telehealth. We'll have to see if those rumors are true. There are some folks out there who believe that the telehealth visit being done at this point in time may be retrospectively covered, but that's something I have not been able to confirm. So I'm going to stay with the recommendation that you get an ABN signed and/or reschedule your patients for those telehealth visits until we see a resolution. If we get anything more on what's happening behind the scenes, we'll certainly let you know.

The CMS directive for the MACs to hold claims for 10 days gives each of the MACs time to adjust their systems so they can process claims once. And it's not just the physician and telehealth that were impacted. Hospital at home is impacted, which has required some hospitals in different states to move people around. So there's more than just the physician payment structure that needs to be addressed by CMS, and the MACs are making those changes. But know that Medicare is funded. It is not part of the budget argument, at least not directly, because those programs are still funded and are mandatory. So this delay is not about funding. It's more an administrative hold so they can get their systems readjusted so they're not paying for telehealth and those types of things. You'll definitely want to prepare for that, but keep submitting your claims to Medicare. The MACs still have people in place. They're just receiving those claims and queuing them up for when they get their revisions in place. So it's at least a 10-day hold. I would count that as 10 business days, but we'll see if it actually is the 10th or it pushes into the following week by a couple of days.

Scott: Ray, comments, thoughts?

Dr. Ray Painter: Nothing to add.

[05:00] Scott: All right. Stay tuned. If you did not receive our email blast announcing the CMS position, you can sign up for those if you go to prsnetwork.com, and on the homepage you can put yourself on our email list. We will keep you informed as we know more. If there's anything to blast out, we'll let you know via email in between podcasts.

What we heard about AI at the section meetings

Scott: On to our next topic. Mark, you and I have been traveling around to a lot of the section meetings and other meetings, and we have attended sessions on AI for various things, and they all seem to be pretty well attended. So first let's get your perspective, then I'll give my perspective, and then we want to get Ray's perspective. What's your perspective on AI?

[06:00] Mark: The common theme is AI is coming, and it is coming in a lot of different directions from a lot of different entities. From one of the talks we went to, and certainly from talking to the attendees and listening to their comments, it really looks like AI is being used or experimented with in multiple areas of healthcare and the urology practice. We heard about AI phone calls for scheduling. We heard about folks who are using ambient listening for recording patient visits with recommended coding. We've seen a number of tools being implemented to scrub claims, recommendations on how to treat, AI enhancements on how to perform procedures. So it really runs the gamut of services within urology.

One of the things I came away with was, number one, it looks to me like AI will evolve in silos to a certain degree, with maybe an AI putting things back together. I liked the general thought that the human part of this is really important in the training of AI. People with certain areas of expertise need to make sure that the AI tools being rolled out and implemented have human supervision. I use the word tool because that's where I've landed on all this. I don't like the name AI as much as I like "the enhanced ability of the computer to wade through data and produce repetitive tasks." There's a lot more information behind what an AI-enhanced protocol is, but in many ways it's a protocol, so it really has to be dialed in to what you're using it for and tweaked specifically for your purposes. That means you're going to have to work with AI to fine-tune AI, and make sure that your partners who bring this into your practice are also doing the same thing.

The computer learns, but it learns in a very linear fashion. So if there's something wrong as it's building its full process, that could actually derail something down the road. It really is like building a house, constructing that process. Paying attention top to bottom is going to be important, and ultimately you're probably going to end up working with a number of different partners depending on what you're trying to do. So you want to make sure that the partners delivering AI to you have a real focus on that expertise. Not just computer programming expertise, but actual expertise and knowledge to understand what you're trying to do. We need people who understand the big picture so they can help drive that, as well as those who understand the steps required to get to that end game.

Buyer beware: a clean claim is not the same as a correctly coded claim

[10:00] Scott: Mark, I agree with you. From what we heard and what we saw, I think the silos is a very good analogy. There are so many possibilities with AI in healthcare administration, to relieve the administrative burden. It's mind boggling. The problem, and you alluded to this, is it takes time to build that tool in a real-life scenario. You can have the best model and build the best prompt, but it isn't until it gets in there and gets a lot of use that it can be refined to the point where it's working toward the version that's really capable of saving you time and energy and producing the right information.

I'll reiterate that there are so many AI opportunities and companies out there approaching urologists and urology practices, and they're touting some significant savings in time and dollars. It's the wild west of AI right now, and trying to figure out the best company to go with is truly buyer beware. Make sure you do your research before you jump in, and really dig into the numbers. An example I'll give: one of the AI companies I saw was boasting, from a coding perspective, that it had a 99% correct claim rate. Well, the problem is that you might be able to feed it a clean claim, but that doesn't mean it was coded correctly from what was actually done in the first place. There was a disconnect in how they approached that. This was a company that was mainly an AI company, and they didn't have a lot of coding expertise on their staff. So when they said, yes, we're getting all these claims through correctly, that's probably true, because AI is pretty good at making a claim go through. But that doesn't mean it was actually coded correctly. There's a big difference. That is where the human needs to be in the loop, really verifying all these algorithms.

Coding is a really gray area, and that takes time to teach AI, because with the gray area in coding and reimbursement, a lot of the historical claims that have been paid may not have been right in the first place. So you may be feeding improper algorithms into the AI model, the large language model. There's a cautionary word here. Just make sure when you're looking at all these things that you're asking the right questions, because the savings numbers they were boasting were pretty impressive. Don't get blinded by that. Make sure you do your due diligence.

[14:00] Mark: One of the things you're leaning into, and one of the things we're seeing, is understand the limitations of the system. In some of these cases we're looking at AI-generated claims, using ambient listening and those types of things, and visits are going from fours down to twos, and they're not understanding why that's happening. A piece of it is understanding that the ambient listening AI only hears what you're saying and what the patient's saying. It's not seeing anything else you're doing: clicking orders, reviewing data. So understanding the limitations, where that information is, how it's processed, what it doesn't have access to and what that particular tool does have access to, will make a difference. Understand what box it's working in and how you need to interact with that box before it can actually be a tool that, A, saves you time, and B, doesn't cost you a lot of money because you missed a bunch of stuff.

Ray's perspective: AI in the loop still needs a knowledgeable brain

[15:00] Scott: Very true. Ray, what are your thoughts? You're muted.

[16:00] Ray: We've always said that AI was coming, but we thought you would need a human in the loop. Well, the way it's developing, and you're talking about the silos and everything, I've now turned it around and said, AI's coming, and it's going to be AI in the loop: AI doing one thing and then AI doing another thing to help move things along. But what I haven't seen is anything that would convince me we don't still need the expertise and a basic knowledge, because you've mentioned that this whole system is detailed, but also the whole coding system is siloed. You've got the doc who does the documentation, and that's what starts it all. Then you've got to have the interpretation of that and the coding of that, and on down the line. Certain parts of that AI can do, other parts not. You're still going to have to have a brain in the loop that's somewhat knowledgeable of the system.

[17:00] Scott: That is important. As we move forward and as you're evaluating all those things coming at you, making sure that you have that expertise behind it, and are monitoring it to make sure it's doing things correctly, does take that knowledge and that expertise.

Sponsor and final thoughts: experiment, but monitor

Scott: Let's wind this episode up and get some final thoughts. We want to thank ModMed for supporting this episode. If you're in the market for an EHR or a practice management system, you can go to modmed.com/prs for specials for our listening audience. Mark, before you give your final thoughts, or while you're giving your final thoughts, can you give a quick summary of the AI discussion we just had? I know we gave a lot of thoughts out there, but can you summarize that in a few bullet points?

[18:00] Mark: Let's try. With AI, I would say go in with your eyes wide open. Don't turn it on and let it run. Make sure you understand what it does and where it fits, and leave your expertise in place so you can check the results of AI along the way. Feel free to experiment with it. I think everybody's going to need to jump in and start touching this as you go through. And I think that even when we have a system that works really well for one practice, it's going to need to be tweaked for the next practice. I can tell you from all of our revenue cycle management work, if you've solved it for one practice, you've solved the issue for one practice. There are always little idiosyncrasies in the way the systems are set up, how they interface with one another, how your staff interfaces with the system, and who's delegated to do what. All of those things need to be taken into account. So, experiment, yes, but monitor.

Scott: That's a good summary. Any other final thoughts you'd like to add?

[19:00] Mark: One: we'll keep an eye on the government shutdown. Hopefully they start talking to one another and getting this resolved, and we'll let you know as we start to see Medicare release their hold. We'll keep you apprised of all of those issues, and certainly keep our ear to the ground to see if we're hearing anything we can let you know about as soon as we can. Ray, final thoughts?

Ray: Well, go back to W.C. Fields. He said, trust everybody, but cut the cards. So when you hear something that's too good to be true, it probably is. Even with AI coming, be careful.

[20:00] Scott: Before we go, I'd like to remind you that we have registration open for the Urology Advanced Coding and Reimbursement Seminars in Las Vegas and New Orleans, and we are going to have more discussion on AI there. We encourage you to register for those seminars, come have these conversations with us, and let's see if we can't make our practices the best they can be in 2026 and beyond. I also wanted to bring your attention to the PRS Coding and Reimbursement Hub. If you go to prsnetwork.com/urologyhub, you can explore the information on the hub, which takes individual products and services and gives the coding for them, as well as category coding. We have a lot of tools available, like quick reference guides, calculators and those types of things. We encourage you to check that out, and we will put links on the episode page. If you go to prsnetwork.com/261 you can find links to register for the seminar, get on our email list, and get to the PRS Coding and Reimbursement Hub. That's all we have for today. Thank you all for listening. Take us out, Ray.

Urology Advanced Coding and Reimbursement Seminar

Information and Registration

PRS Coding and Reimbursement Hub

Access the Hub

Free Kidney Stone Coding Calculator

Download Now

PRS Coding Courses

For Urologist

For APPs

For Coders, Billers, and Admins

PRS Billing and Other Services - Book a Call with Mark Painter or 
Marianne Desciose

Click Here to Get More Information and Request a Quote