In a recent HFMA Region 8 webinar, “AI in Denials Management: From Concept to Reality,” Aspirion‘s Jim Bohnsack, Chief Strategy & Client Officer, and Spencer Allee, Chief AI Officer, provided an insightful look into how artificial intelligence (AI) is transforming healthcare denials management. The presentation highlighted how emerging AI technologies are helping providers counter increasingly sophisticated denials strategies from payers.
The Current Denials Landscape
“Nobody would be surprised and we’re all facing the fact that denials are going through the roof,” Bohnsack explained. “And in particular, what we’re noticing is it’s not just the volume of denials, but the flavor of denials that we’re seeing that are also troubling.”
Payers are making substantial investments in payment integrity initiatives, leveraging advanced technologies for claim reviews.
“We’re seeing them invest heavily in technology,” said Bohnsack. “The speed with which they’re applying this technology to review claims is unprecedented and, quite frankly, because they have the technology now, they can do a more comprehensive review in a shorter amount of time of more claims.”
Of particular concern is the rise in complex clinical denials requiring detailed validation. Payers are using machine learning (ML) models to identify claims they believe may be billed at higher complexity levels than warranted, resulting in downgrades and partial denials that require substantial clinical evidence to appeal.
The Economics of Appeals
Historically, Aspirion has relied on attorneys and clinicians to craft appeal letters, leveraging their expertise in analyzing medical records against various guidelines and policies. However, this approach faces significant economic constraints. As Bohnsack explained, while it makes financial sense to dedicate attorney time to appealing a $10,000 claim, the same effort becomes difficult to justify for a $2,000 downgrade—despite requiring the same intensive work.
The AI Solution: Doc IQ
To address these challenges, Aspirion developed Doc IQ, accelerated by their September 2023 acquisition of Infinia ML. Allee explained their technical approach: “We use an approach called RAG (Retrieval Augmented Generation). This system only looks in the four corners of that medical record to extract information. So it’s not co-mingling data and it’s not making up data or making up answers if it can’t find it within that medical record.”
Proven Results
Early testing showed impressive results. As Allee shared, “In our first MVP release, the platform-generated letters slightly outperformed the human generated letters by about 5 percent. We expanded the platform in December and made some tweaks and improvements, and then we did the same blind test and those letters outperformed the humans in those cases by close to 30-35 percent.”
Key Features of Doc IQ
The platform includes two primary modules:
1. Assistant Module: Helps users efficiently search medical records using predefined questions, providing specific page references for relevant information. Users can also ask custom questions, with all queries saved for future reference.
2. Appeals Module: Offers a four-step workflow for appeal generation, including denial type selection, code entry, denial letter upload, and information verification. The system processes information in 3-5 minutes, allowing users to work on other tasks while appeals are generated.
The Technology Edge
Bohnsack emphasized the transformative nature of large language models (LLMs): “The ability for us to take unstructured data, structure it into these large language models, and produce a generative output that has some level of reasoning in it is what has really changed the game in terms of our ability to truly automate a process.”
Beyond Technology: A Human-AI Partnership
The platform represents a strategic response to payer sophistication. As Bohnsack explained, “As we look at the market and what’s happening—the payers’ investment in this technology on the flip side—we’re getting outgunned and outmanned and so we really have no choice but to start to explore technologies like this to kind of balance that equation.”
Aspirion is expanding the platform’s capabilities beyond DRG downgrades to address other denials types, including inpatient-to-outpatient downgrades and various medical necessity denials. This expansion represents a significant investment in the “tens to dozens of millions of dollars” but one that Aspirion views as essential to help providers counter growing payer sophistication in denials.
“We are better than either just machines or just humans alone,” said Allee in describing the quality-assurance process. “We are building this platform that uses our subject matter experts (SMEs) in the places where they’re most valuable reviewing what’s produced here by the platform and then also dealing with more complex cases that we can’t handle yet.”
A Necessary Response to Advanced Payer Strategies
Importantly, Doc IQ remains an internal tool rather than licensed software. Aspirion continues to operate on a contingency basis, receiving payment only when providers successfully recover denied claims. The technology simply allows them to handle more appeals more efficiently, particularly for lower-dollar cases that were previously cost-prohibitive to pursue.
The webinar highlighted how AI is moving from concept to reality in healthcare revenue cycle management. While the investment in such technology is substantial, it represents a necessary response to increasingly sophisticated payer denial strategies. By combining human expertise with AI capabilities, organizations can more effectively address the growing challenge of claims denials while maintaining appeal quality and expanding their ability to pursue appropriate reimbursement across a broader range of cases.
Through Doc IQ, Aspirion demonstrates how AI can transform traditional appeals processes, enabling providers to efficiently challenge denials across a broader range of cases while maintaining high-quality standards in appeal generation.
To learn more, watch the entire on-demand webinar here.
