Optimizing Clinical Denials Management with AI

Summary

---

Podcast Summary: Optimizing Clinical Denials Management with AI

Podcast: Becker’s Healthcare Podcast
Episode Title: Optimizing Clinical Denials Management with AI
Date: September 5, 2025
Host: Lucas Voss (Becker’s Healthcare)
Guest: Jim Bonesack (Chief Strategy and Client Officer, Experian)

---

Episode Overview

In this insightful episode, Lucas Voss talks with Jim Bonesack, an experienced revenue cycle leader, about the pivotal role of artificial intelligence in optimizing clinical denials management. The conversation covers current challenges faced by providers, the transformative impact of AI technologies, and strategic approaches for healthcare organizations to stay competitive amidst growing payer complexity and administrative cost pressures.

---

Key Discussion Points & Insights

1. The Clinical Denials Landscape: Pre- and Post-AI

[01:31 – 05:09]

• Historic Imbalance:
  • Traditionally, payers have broad visibility across many claims but lack clinical depth; providers have deep clinical data on individual claims but lack payers’ breadth.
  • This imbalance leads to ongoing friction and growing numbers of clinical denials.
• Emergence of Complex Denials:
  • Recent years have seen a significant increase in denials requiring detailed clinical appeals, especially for inpatient stays.
  • Payers increasingly deny claims pending clinical validation, shifting from simple coding errors to nuanced medical necessity issues.
  • Providers now face demands to extract, collate, and present justification from complex medical records, requiring skilled (and costly) talent like nurses and coders.
• AI’s Role:
  • Payers are leveraging large language models to audit and automate denials, raising the bar for providers to respond efficiently and comprehensively.

Quote:

"These are, I disagree with you. It wasn’t medically necessary or wasn’t that complex. Write me a letter pulling clinical documentation and evidence and send it to me and prove it. ... That’s tough for providers. They don’t have a bunch of resources sitting around wanting to write letters."
— Jim Bonesack [03:07]

---

2. AI-Driven Optimization: Results and Metrics

[05:09 – 08:59]

• Strategic Approach to Implementation:

  • Experian adopted a phased approach—incrementally deploying AI tools for specific appeal tasks rather than full end-to-end automation.
  • Focused on supplementing human effort, ensuring AI-produced appeals are as good or better than those written manually.

• Impact Metrics:

  • Automated Appeals Coverage: AI now covers almost 70% of clinical denial types for generating automated appeal letters.
  • Speed and Efficiency: Reduced time from placement to appeal writing by 35% (removing about 40 days from the process).
  • Quality: Overturn rates on appeals increased by ~25%, with AI surfacing more comprehensive evidence from medical records.
  • Reduction in Appeal Volume: Fewer total appeals needed, as initial submissions are more robust (10-15% reduction in subsequent appeals).

Quote:

"We’ve lowered the days from our placement to appeal writing by 35%. So we’ve taken 40 days out of the process of getting an appeal out the door. ... The overturn rate going up beyond what a human can do is also super important."
— Jim Bonesack [06:35]

---

3. Strategic Advice for Healthcare Leaders

[09:00 – 13:04]

• Challenges Beyond Technology:

  • Increasing payer tactics (e.g., post-payment reviews, DRG downgrades) are intensifying the workload.
  • Acknowledges some payer strategies as “dirty pool,” emphasizing the unfairness and complexity of current practices.

• Not a Human-Scalable Problem:

  • The demand for clinical staff to fight denials is unsustainable and often unfulfilling.
  • Recruiting and retaining skilled workers for this task is increasingly challenging and costly.

• Call to Action for Provider Organizations:

  • AI and technology are necessary to respond to denials at scale—adding more staff is no longer viable.
  • Providers should honestly assess their resources and consider:
    • Licensing technology to empower internal teams
    • Partnering for full-service solutions
    • Determining the right blend of tech and people for their situation

Quote:

"It really is. It’s not only volume of denials, but it’s the particular type of denial and the response required to address it. ... I’m convinced ... it’s not a human scale problem. We can’t add more people and fix this problem. We have to respond in a technological way."
— Jim Bonesack [11:58]

---

4. Final Thoughts and the Future

[13:39 – 14:51]

• Define “AI” Clearly:
  • There’s a lot of industry hype; leaders must understand what AI/ML means in practical terms and how it can be implemented.
• Closing Gaps:
  • Payers are outpacing providers in AI adoption, requiring urgent action from provider organizations to close the gap.
• Multiple Paths:
  • Organizations should evaluate different AI solutions; there’s no one-size-fits-all.
  • The pace of change in healthcare administration is unprecedented, offering both challenge and opportunity.

Quote:

"Health care can’t afford to keep this pace. So we, we have to do it. How do we do it? In a thoughtful way. Multiple approaches. But we’re all getting towards the same thing. It’s tremendous. In my career I’ve never seen the pace of change this fast."
— Jim Bonesack [14:25]

---

Notable Quotes & Memorable Moments

• On Administrative Complexity:
  "Whereas we used to get lesser amounts of clinical denials ... Those are growing exponentially at this point, where the volume is just out of control."
  — Jim Bonesack [02:27]

• On Provider Challenges:
  "I don’t know many coders or nurses that say I went and got trained and certified ... so I could write letters back to the payer to argue something. ... That’s not a satisfying job."
  — Jim Bonesack [12:17]

• Sports Metaphor on Appeals:
  "Three shots on goal, but you got to stay on the ball, right?"
  — Lucas Voss [14:51]

---

Timestamps for Important Segments

• 00:34 — Jim Bonesack’s career background and expertise
• 01:31 — Current and historical challenges in clinical denials
• 05:44 — Concrete examples of AI-driven outcomes
• 06:35 — Quantifiable impact of AI: speed, quality, and volume metrics
• 09:40 — Provider skepticism and strategic advice
• 11:58 — The necessity of a technology-driven solution
• 13:39 — Final thoughts on defining and implementing AI
• 14:25 — The unprecedented pace of industry change

---

Conclusion

This episode underscores the urgent need for healthcare providers to embrace AI and advanced technology solutions to manage spiraling clinical denials, reduce administrative burden, and optimize reimbursement. With practical metrics, candid commentary, and an honest assessment of industry dynamics, Jim Bonesack offers clear guidance for leaders seeking both immediate ROI and sustainable, long-term advantage.

---

Transcript

[A]

Hi everyone, this is Lucas Voss with Becker's Healthcare. Thanks so much for tuning in to the Becker's Healthcare podcast series. Fantastic to have you. An exciting topic today, AI's role in clinical denials management. And I'm very pleased to be joined by Jim Bonesack, Chief strategy and Client Officer at Experian. Jim, thanks so much for being here today. It's great to have you.

[B]

Yeah, it's good to be back. Thanks, Lucas.

[A]

Yeah, great to have you. For our audience that might not know you, they absolutely should, but they may not. Could you share a little bit about yourself and your work in healthcare?

[B]

Yeah, I've been doing this for 25 plus years now at this point and early career was consulting with Ernst and Young and Deloitte doing turnarounds for hospitals, plus epic implementation at Kaiser, things like that. And then I shifted into strategy, product M and a and spent two different stints at TransUnion Healthcare, one at Conifer, organizations like that, all focused on revcycle, a lot of M and A and product strategy. What do we build, what do we buy? That kind of focus. And for the last two and a half years I've been here in this role where I have a very similar focus on what do we, where are we going, how are we helping providers? Do we build it, do we buy it, how do we work with our current clients and our future clients and kind of how does the business evolve to meet the needs of the market. So it's a fun, it's a fun job. Yeah.

[A]

And we'll certainly talk about some of this stuff and I know you're very familiar with the numbers around this topic.

[B]

Right.

[A]

And denials continue to cost healthcare organizations ability. Yes, that's with a B billions. And AI is certainly something that's playing a critical role in this process and how that process is evolving. Can you walk us through a little bit of what the clinical denial landscape looks like right now or and looks like before implementing AI? What did it look like before?

[B]

Yes, I mean the evolution and the changes that are happening in the market right now based on broadly advancements in technology is crazy. And it's crazy from both the payer and provider side. And you can see the impacts that are going on. Both, you know, clinical denials themselves have been around for my whole career. I think they've changed materially in kind of flavor and type, especially in the last, call it 18, 24 months. You know that I think the, the denials themselves, kind of an imbalance in, in information between payers and providers has always caused level of friction where the payers see the landscape broadly across all the claims that are received for a member. And so they've got this very horizontal view of what's going on in the spend and cost and utilization. But they don't have visibility deeply into any given visit or claim and especially in the inpatient setting where they tend to be the most expensive. Whereas the providers obviously have it very deep in their clinical setting but they don't have it as wide across. So that imbalance has created this tension between the two forever. And so in order to protect against what the payers would say is bad medicine, they would say I need you to get approval through the form of an authorization or pre certification or other and send me some level of justification. This is medically necessary and a good use of medicine and resources. So authorizations and all the noise around authorizations were super. You know, it's still a friction point today I'd say it's becoming less so as the payers shift their technology. It seems to more of a payment and integrity back end process of reviewing clinical documentation. And so whereas we used to get lesser amounts of clinical denials that required an appeal and us to justify medical necessity for something we did, those are growing exponentially at this point. Yeah. Where the volume is just out of control and it's not as easy as your old I would call it. We grouped them as clinical denials. They were more coding related like don't use this modifier, this diagnosis and procedure code don't go together. Some of those errors where simpler to fix. From an administrative perspective. These are, I disagree with you. It wasn't medically necessary or wasn't that complex. Write me a letter pulling clinical documentation and evidence and send it to me and prove it. Yeah, okay. That's not your normal business office PFS person that can respond to that. That requires nurses or coders or other. And so somebody with a high degree, that's expensive and those kind of things to pull medical records to justify that. The reality is I think what they're doing is denying for medical records, loading it into large language models and then producing a response that says we disagree. We segmented this data that was historically unstructured and not available to me and pushed it back on you to say I disagree. Prove me wrong. That's tough. It's tough for providers. They don't have a bunch of resources sitting around wanting to write letters. So it's a challenge for sure.

[A]

You've talked a lot about optimization when it comes to this process. Right. And you just walked us through exactly what the problem is. It needs optimization, it needs somebody looking at it in terms of where can we make it better, where can we improve. Right. And you said it's really a critical step to stabilize margins, being able to optimize all of this. What ROI have you seen from AI in clinical denials management right now, today? And what's an example from your work, from your recent work that stood out from an implementation that really made a difference for you?

[B]

Yeah, we've been on this journey, you've known since we talked, but we've been on this for over two years. An investment and bought a coal company and dedicated number of 35 people working on this full time. And specifically working on this is not a human addressable problem for the providers or for Asperian where we're getting this huge volume of clinical denials that require clinical validation. I have to pull a medical record, I have to review that medical record to find evidence against a claim that says, I disagree with you and what are the rules that apply. So is it care guidelines, coding guidelines, is it payer policies, is it managed care contracts? So all of those things are super complex and you got to pull all those things in together. And so as we've been on that journey, we have kind of gone through it very systemically and saying, all right, we're going to produce some of these AI applications to our end users just to make them a little more efficient in their job and release them piece by piece versus too many people make the mistake, I got to automate this thing end to end before I release it or it's not going to work. And that's not true. So stepping through it, we were seeing efficiency gains and just our throughput today we cover almost 70% of clinical denial types with an automated appeal letter generation capability. And what we've really seen in terms of metrics which are very attributable to our providers is we've lowered the days from our placement to appeal writing by 35%. So we've taken 40 days out of the process of getting an appeal out the door. We've seen an increase in the overturn rate and successful Appeals by about 25%. So, and it's not because our humans weren't good at it, it's because the machine doesn't miss all the evidence within a 500 page medical record. Right. To write a very compelling argument and share that information back with the payer to say, no, no, no, this was clinically necessary. And then in addition we've seen a 10 to 15% lowering in the number of appeals required. So you know, there's a first level appeal, second level appeal, third level appeal. You basically have three shots on goal to argue your case. That's dropped by between 10 and 15% the number of appeals. So also pointing to the fact that it's taking less because they're more comprehensive and more effective in overturning by providing the evidence, the clear evidence. And so those are exciting because they're, you know, 40 days off days in AR in this bucket is huge for providers. And the overturn rate going up beyond what a human can do is also super important. That was our, our starting point. It's like this needs to be as good or better than what a human can do and certainly we know it's going to be more efficient and faster. Got it. But can it, is the quality as good? That was our key principle and it's proven to be far better. So that's awesome. Yeah.

[A]

So again that's what, when we talk about optimizing, it's really the efficiency component that you've mentioned that's really key, creating more efficiency. But I love that you mentioned it's also about quality. Right? Yes, we want to be more efficient, but the quality needs to be there, which is really, really key.

[B]

Right.

[A]

You again, you work in this space every day. You see the numbers. We started our conversation with that too. I'm really interested in if there's anything that surprises you about the space about AI, about denials. When you work in this space, when you have these conversations, is there anything that pops up that surprises you? And what do you tell leaders? What do you tell hospital leaders, healthcare leaders when you have those conversations that are, that are still on the fence. Right. That are like hey, I don't really, I believe in my people, I know that they can do everything. I don't necessarily need this. What type of conversations do you have? What do you tell them?

[B]

Yeah, it's nothing. After 25 plus years in this space surprises me. It they had the amount of changes, the multi dimensional nature, the 4D chess of this environment with payers, providers, you know, patients, members, all playing, employers, playing roles and in how the changes are occurring across the market. Kind of nothing surprises me. I will be, I am surprised at times by some tactics and some kind of games that the payers play in issuing these denial times. I mean we're seeing them post payment, review, audit, those are kind of common. We request a medical record issue a denial letter, say we disagree here's why there's a couple out there that are getting a certain bill DRG and then just paying a downgraded DRG and putting a payment adjustment reason code on it. That's contractual adjustment. Like that's to me that's dirty pool. Like you're saying, this was that complex and I'm expecting this amount of reimbursement because I exerted this amount of effort. You're not even getting the clinical documentation. You're just saying disagree, I'm going to downgrade it, pay you less and see if you fight it like that. That not surprising, but it's dirty, right? It's not very fair, but all leading to the same problem. And so as it relates to our provider clients and colleagues, it's like I think this onslaught's going to continue. I think the financial situation that the payers just reported the last couple of weeks is going to get them to crank that denial dial up and put more focus on paying less. Their alternative is go sign up more members or go raise rates. Well, that's not palatable for anybody. So I think, you know, one of the ways is to get more efficient or pay out less, you know those two levers and I think they probably go with payout less. So I think this volume problem is going to continue. And it really is. It's not only volume of denials, but it's the particular type of denial and the response required to address it. So it is that clinical validation. I need a coder, I need a nurse. Well, I don't know many coders or nurses that say I went and got trained and certified and all of these things so I could write letters back to the payer to argue something like that's not a satisfying job. So finding them one, hiring them and the expense associated with them, plus having them want to do that job, that's a lot of things that the providers are challenged with when it's this type of denial. And so, and as we looked at that, it was very similar internal to us as we got to try to hire and keep those people and have them do this job. So it is really a technology, it needs a technology solution. I'm convinced of that. It's not a human scale problem. We can't add more people and fix this problem. We have to respond in a technological way. There's a bunch of different ways you can license tech and if you've got the people, you can do a full service like us where we use tech and the people, you can just do the service and Throw bodies at it if you try. So I think providers need to figure out what their capabilities are, what their capacity is to deal with it internally or not and should they license something to empower my people to be more efficient or use a provider partner payer to help do this like us or what alternatives. But you have to address it. It's going to get worse and I'm worried that it's going to hurt the provider community more so than we're already not in a great financial situation.

[A]

Yeah, it's a key to long term strategy. I feel like if you're not short term, obviously fixing some of the issues that we talked about but then also long term being able to address this in the future to make it better for everybody else that's coming. Right. And sort of the future piece of this. Or you can simply have a conversation with Jim which will fix that problem, I think.

[B]

Yes, I'm not so sure about that, but happy to share what I've learned.

[A]

Jim, it's so great to have you again. Thanks for being here. I want to turn the floor over to you. Anything else that you want to mention to our audience? Any final thoughts that you have from our conversation today?

[B]

I think it's confusing and there's a lot of hype and this buzzword AI ML and all this stuff gets thrown out broadly and clearly defining what it is is super important and how you're using it and how you're deploying it. These large language models in particular are changing the game. And I think the provider community is getting outpaced by the payer community, no doubt. And so again, incumbent upon folks like us to help develop those type of solutions to even that playing field. And so, you know, we're taking one approach where we're building it internal and using it for efficiency and then still delivering a result. There are others that are building the technology and licensing it out. There's epic. That's saying we've got some AI embedded in our workflow and appeal writing and other, you know, again, not one size fits all. But we have to drive some of this cost and inefficiency out of the system. That's administrative waste. Health care can't afford to keep this pace. So we, we have to do it. How do we do it? In a thoughtful way. Multiple approaches. But we're all getting towards the same thing. It's tremendous. In my career I've never seen the pace of change this fast. So it's super exciting to, to be a part of it. Yeah.

[A]

To come back to your sports metaphor. Three shots on goal, but you got to stay on the ball, right?

[B]

100. Yeah, you got to.

[A]

You got to stay right there. Well, Jim, again, thank you for your time for being here. It's so great to have you. And we also want to thank our podcast sponsor, Esperiana. You can tune in Tom podcast from Becker's Healthcare by visiting our podcast page at beckershospitalreview.

[B]

Com.