RISE National 2025: Fathom panel on autonomous risk adjustment coding

At RISE National in San Antonio (March 11-14, 2025), Andrew Lockhart, CEO of Fathom, and RaeAnn Grossman, CEO of HLTHworks and Medicare Advantage expert, led a session exploring the evolution of technology to serve health plans' changing needs. This recap synthesizes key takeaways and actionable advice from the discussion.

Key takeaways

1. Market pressure demands new approaches. With rising Medicare Advantage enrollment and heightening regulatory scrutiny, traditional coding approaches can't scale efficiently while meeting accuracy expectations.
2. Technology has crossed a threshold. Deep learning AI advances have enabled 90%+ chart-level automation at higher accuracy rates than manual coding, fundamentally changing the cost-quality equation for risk-adjustment operations.
3. Early movers are gaining lasting advantages. Health plans and risk-based providers can accelerate technological adoption with a risk-free proof of concept that forces the vendor to prove performance at scale before any commitments, while strong contractual SLAs ensure ROI realization during production.
‍

Summary

1. Industry updates

Market conditions forcing operational changes. Risk adjustment is more challenging than ever, with Grossman noting that "RA seems to be harder and harder." Medicare Advantage's explosive growth, coupled with the zero-sum nature of ACA risk adjustment, has created what she called an "ICD-capture arms race" – in which plans must continuously invest more resources just to maintain their position relative to competitors.

Compliance stakes are higher. Regulatory scrutiny continues to intensify, demanding both greater accuracy and more robust validation processes. Policy volatility from the new administration may impact reimbursement schemes. In this fluid environment, organizations increasingly need solutions flexible enough to adapt quickly to changing rules.

Quality at scale is the central challenge. As chart volumes grow, maintaining consistent quality becomes difficult with traditional coding methods. RA leaders increasingly need completeness and high specificity of coding to achieve both compliance and financial goals.
‍

2. Technology evolution

To support risk-adjustment coding, technology has developed in three main phases.

2000s to mid-2010s: Early NLP-based tools show promise. These technologies enable automation for simple coding scenarios, but struggle with the complexity and chart length involved in risk-adjustment coding.

2015-2018: ICD-10 creates major disruption. The ICD-10 transition in 2015 increases the number of codes tenfold, causing automation to fall off a cliff – even for basic cases. Organizations begin limiting technology usage due to poor coding quality and compliance risk.

2018 onward: Deep learning transforms capabilities. First end-to-end automation of coding arrives in the market, starting with provider-side coding before expanding to cover RA for payers in 2022. This approach unlocks 90%+ automation at high accuracy.
‍

3. Implementation considerations

Differentiate autonomous coding from coding assistance. The vendor market includes both fully autonomous solutions and legacy computer-assisted coding tools. True autonomous systems code full charts without human intervention, while assisted tools merely suggest codes requiring manual review. Only truly autonomous technology delivers transformative ROI.

Select vendors based on adaptability and team strength, not just price. "Customer centricity is most important. Cost is always the early winner, but [choose] someone who will actually take the time to match your own workflows and listen to how you want your data integrated," Grossman advised. Vendor team strength directly impacts client value realization.

‍Define target outcomes and back them with strong SLAs. Before evaluating technologies, clearly establish your priorities: coding accuracy, RAF capture rates, cost savings, turnaround times, or some combination. These specific metrics should then be reflected in contractual SLAs. As Lockhart cautioned, "If vendors are waffling about an outcome metric going into the contract, in all likelihood it won't go in and you won't get that outcome."

Run a high-volume proof of concept to validate performance. Lockhart emphasized the importance of testing autonomous coding at scale before full implementation. "What we do is code an enormous volume upfront. Then the client can evaluate it and know it's going to work at scale." This approach proves out capabilities and builds internal stakeholder confidence.
‍

4. Impact and outcomes

Early adopter demonstrates tangible returns. During the session, Lockhart shared results from a health plan with approximately 2.3 million members that achieved 91.2% successful chart-level automation without human intervention. This resulted in a 48% reduction in vendor coding spend while improving quality – shown by a 43% reduction in coding errors, 37% increase in ICD capture, and 42% increase in RAF scores.

Rapid turnarounds unlock far-reaching benefits. "When everything is coming back within a day, think about what that does for your organization," Lockhart noted. RA operations fundamentally shift with 24-hour or faster coding turnarounds. Within the same submission window, teams can now analyze results, correct issues, iterate, and refine their approach.

Consistent AI eliminates coding variability. The panelists highlighted how AI applies coding guidelines consistently across all charts, unlike manual processes prone to coder fatigue and interpretation differences. Lockhart emphasized that AI's exhaustive analysis of every chart leads to higher net gains than initially anticipated and establishes a higher quality baseline.
‍

5. Strategic implications

‍Teams evolve toward higher-value activities. As routine coding becomes automated, staff roles shift toward strategic analysis, quality oversight, and cross-functional collaboration. Lockhart described the opportunity for RA leaders: "Imagine a world where you had 1,000 McKinsey consultants who knew everything about your organization. We're trending toward a world of abundant intellectual capacity."

‍Early movers gain lasting advantage. "The clinical side [for AI] will take longer because of regulations, but the admin side is less encumbered," Lockhart explained. This gap enables early adopters to build significantly more efficient administrative operations while competitors rely on manual processes. Lockhart emphasized that organizations moving quickly now will be better equipped to implement AI across other functions over time, enhancing their position.
‍

If you'd like to learn more about Fathom, schedule a meeting here.