The following recaps a Spark Session fireside chat delivered by Aspirion’s Chase McGrath, Vice President of Product, and Rikki Ashkin, JD, Senior Client Success Director, at the HFMA Region 8 Mid-America Summer Institute (MASI) in Des Moines, Iowa. The session paired two vantage points on the same problem—operations and technology—to map what a smarter, more proactive denials strategy actually looks like.
Ask most revenue cycle leaders whether denials are getting easier, and you’ll get a tired laugh. The data explains why. In 2025, 41% of providers reported denial rates of 10% or higher. And it isn’t just frequency—the dollars behind each denial are climbing too. The average amount of denied inpatient claims rose more than 12%, and denied outpatient claims more than 14%.
The through-line: payers are using automation and AI to deny claims faster, at higher dollar values, and with more sophistication than ever. The playing field has shifted—and as Ashkin and McGrath framed it in Des Moines, reactive strategies simply aren’t enough anymore.
What made their MASI session different was the format—not a lecture, but a conversation between two lenses. Ashkin brings the frontline operations view: the appeals piling up, the deadlines that don’t move, what it takes to fight and win a case. McGrath brings the builder’s view: how AI is engineered and deployed against denials, and how to separate what’s genuinely working from what’s just hype. Here’s where their perspectives converged.
What Actually Moves the Needle
The first question on the table: how are peer health systems structuring denials programs that produce real, measurable results—not just another dashboard?
From Ashkin’s frontline vantage, the real work starts upstream. The programs that move the needle address documentation, authorization, and coding at the front end—not just appeals at the back. And they assign clear ownership for root-cause fixes, rather than celebrating case-by-case wins while the same denials keep recurring. Her clinical-legal experience adds another layer: recovery is maximized when payers are handed the full clinical and contractual picture, not just an appeal that happens to beat a filing deadline.
McGrath’s build-side answer complements that. Getting ahead of denials, he argued, requires pattern intelligence—denial data segmented by payer, denial type, and DRG, then tracked over time so payer-specific behavior signals become visible. Those signals feed a post-resolution feedback loop that actually drives upstream action. The critical distinction: the best programs connect that intelligence directly to a fix. A dashboard that surfaces a trend but never changes what happens before the claim goes out is just tracking. It never turns into action.
That gap has a cost, and the session made it concrete. Consider a payer that quietly changes its chemotherapy authorization policy. Denials start immediately—but the new requirement never reaches front-end staff. In the historical loop, claims get denied, appealed 60 to 90 days later, and leadership only spots the trend once the financial impact has mounted. Root cause found, staff finally educated—roughly 180 days later.
In an AI-enabled model, the policy change is flagged, front-end workflows are updated before services are rendered, cleaner claims go out the door, and every denial that does land is automatically root-caused for rapid trending. Same problem, resolved before it ever hits the AR.
Cutting Through the AI Hype
If getting ahead of denials is the goal, AI is the tool everyone’s selling. So, the second conversation tackled the hype directly: where are teams seeing real results, and where is it still just noise?
McGrath was candid about the overpromises. The claim that AI can predict denials before they happen? Insights from denials are inherently ad hoc—but AI can still shorten the CDI and utilization management feedback loop and help close process gaps faster. The idea that AI only helps with appeals? In practice, it runs across the end-to-end process—reading payer correspondence, checking claim and appeal statuses, and more. The promise of significant value right out of the box? Real lift shows up after a learning curve, for both the technology and the team, and RCM impact takes time to demonstrate.
And the biggest misconception: that AI replaces clinical and legal expertise. It doesn’t. While AI assists on the simpler cases and frees experts to make the human call on the ambiguous ones, AI templates were developed, validated, and monitored by both legal and clinical subject matter experts.
It’s the reason both presenters were wary of a claim you’ll hear on any exhibit floor—that a system has “automated appeals with just generative AI embedded in its EHR.” Complex clinical denials demand more than a generative model bolted onto a chart. The key message is that while AI is extremely capable of handling more straightforward denials, it does so because it has been built on expert knowledge, not in place of it.
Proof the Model Works
The conversation wasn’t theoretical. Two client examples anchored it.
In the first example, a provider’s clinical denials were piling up across every payer faster than the team could work them—rising AR, delayed reimbursement, real financial pressure. Routing every case for AI-generated, payer-specific appeals—each one AI-drafted, then human-verified before submission—recovered more than $5.6 million in two months, with 86% of collections coming from first-level appeals and no additional internal staff required.
In the second, the target was write-offs. Auto-denials are increasingly used to test whether a provider has the resources and the will to appeal, and rural and resource-constrained hospitals are common targets—often defaulting to write-offs even on cases that clearly meet inpatient level-of-care criteria. By routing 100% of eligible cases for appeal, with no cherry-picking, the provider reached a 75% overturn rate on inpatient-level-of-care denials and $1.3 million in quantified recoveries—while signaling payers that write-offs won’t go unchallenged.
From Denials Data to a Decision
The final conversation addressed the part that determines whether any of this gets funded: translating denials data into a story that leadership will act on.
Ashkin and McGrath laid out a four-step path. Capture the pattern—segment denials by payer, type, and root cause, not just a blended rate. Quantify the dollars—price out what each pattern costs, not just how often it happens. Build the narrative—turn the data into a plain story leadership can act on, not a spreadsheet they’ll skim. And drive the decision—use that evidence to justify strategy, staffing, or even contract renegotiation.
Denials data, told right, isn’t an operations metric. It’s a strategic conversation.
An Operational Shift, Not Just a Technological One
The session closed on a reframe worth sitting with. AI already speeds appeals and surfaces payer patterns—but its potential runs across the entire back-end revenue cycle. The teams pulling ahead treat denials data as an input to a tech-driven denials strategy, not just a set of operational metrics. Pattern intelligence is what breaks the reactive trap—the mechanism that turns chasing denials into getting ahead of them.
And the most important point: adopting AI for denials is an operational shift, not merely a technological one. The tool matters, but so does the willingness to rewire ownership, feedback loops, and front-end workflows around what the data reveals. Payers have already made that shift. The question Ashkin and McGrath left the room with is the one in the session title: they’re getting smarter—are you?
To learn more about building a smarter, AI-enabled denials strategy, connect with the Aspirion team.
