In a recent interview with the AI Catalyst team, a leading revenue cycle executive shared how an AI-powered CDI tool, Iodine’s AwareCDI, has transformed clinical documentation integrity (CDI) within their 20+ hospital health system.

The impact has been significant: By improving query rates and accuracy of claims coding, the health system achieved $15 million in additional revenue in its first year of implementation.

(Editor’s note: The health system requested not to be identified due to an organizational policy against discussing vendor relationships.)

The challenge: Health system CDI staff was overburdened, leading to big missed opportunities

Accurate and complete clinical documentation ensures proper reimbursement, supports quality reporting, and provides a true picture of patient care. However, many health systems struggle with inefficient CDI processes that leave potential revenue uncaptured.

At this particular health system, the status quo approach involved CDI staff combing through all coded patient notes to identify those in need of additional documentation. This process was time-consuming and often inefficient, limiting the number of claims that could be queried. As a result, revenue was potentially left on the table, and opportunities for documentation improvement were missed.

The challenge was clear: How to improve CDI efficiency without overburdening staff or disengaging physicians?

The AI solution: Intelligent prioritization of high-value claims

To address these challenges, this leading health system implemented Iodine’s AwareCDI, which uses machine learning and healthcare-specific large language models. AwareCDI works by recognizing patterns in historical patient notes which were successfully queried. It then assigns a score to each patient note within a CDI employee's workpile that reflects the likelihood and impact of a query.

The health system’s goal was not to reduce employee headcount (FTEs), but rather to help overburdened staff make better use of their time by autonomously prioritizing high-value workstreams. As their rev cycle VP noted, "If anything, I could increase [FTEs] now because of how much more stuff is getting prioritized."

Outcomes and lessons learned

By improving query rates and increasing accuracy of claims coding, the AI solution led to an additional $15 million in revenue in the first year of implementation. The health system’s overall query rate rose from 20% to 37% of all claims submissions.

Interestingly, while query rates have increased significantly, physician productivity has also improved. Hess observed that physicians are now "working both smarter and harder" due to the AI's streamlining capabilities – but without feeling burned out or disengaged. Physicians are spending the same amount of time on queries but are now able to work at twice the previous rate, with 90% of query responses coming within 24-48 hours or less, and the remainder coming within 72 hours.

This health system’s experience offers several key insights for health systems considering AI implementation in revenue cycle management:

AI-powered CDI tools have diminishing returns – requiring adaptive implementation strategies. While AwareCDI drove $15 million in increased revenue in the health system’s first year, in the second year, it drove only an additional $7 million. This wasn’t due to any decrease in solution performance, but stems from the fact that process improvement in revenue cycle has diminishing returns as your shop becomes more efficient. In the case of CDI, documentation improvement can often be achieved at the outset of the revenue cycle process, which reduces the query rate. Additionally, even when the query rate increases, many flagged claims will be highly complex, whereas fewer lower-complexity claims will be flagged. This instance of diminishing returns highlights the need for health systems to set realistic long-term expectations and prepare for possible fluctuations. It also underscores the importance of viewing AI as an ongoing investment rather than a one-time solution, requiring sustained efforts to maintain and improve performance over time.

AI tools can spark continuous physician education and improvement. The health system’s experience with AwareCDI highlights the decisive role that dedicated CDI educators can play in boosting rev cycle operations. Their six-person CDI education team leverage AwareCDI to ensure physicians understand and effectively engage with CDI processes. By flagging queries in a more organized, bundled manner, AwareCDI enables the education team to send effective best-practice alerts in a way that highlights overarching trends and documentation mistakes. The goal shifts from merely addressing current documentation issues to proactively preventing future mistakes.

Explainable AI outputs are crucial for fostering employee trust. Importantly, AwareCDI justifies its recommendations with specific evidence from patient documentation, highlighting specific problematic sections within the patient's chart or outlining missing information accompanying certain diagnoses. This evidence-based approach allows physicians to quickly verify the AI's reasoning, making the technology feel like a knowledgeable colleague rather than a black box. What’s more, the information surfaced through AwareCDI’s explainability functionality feeds into Iodine’s monthly meetings with the health system, allowing both parties to continuously refine their AI model.

Questions to consider:

1. How can your organization prepare for the potential increase in high-value work that AI implementation might bring, and how might this impact staffing strategies?

2. What steps can you take to ensure physicians and staff view AI as a productivity-enhancing tool rather than a threat to their roles?

3. How can you measure and communicate the success of AI implementation beyond revenue increases, particularly in terms of workflow efficiency and job satisfaction?

The Epic Dilemma: Should You Gamble on Epic’s AI Ecosystem – or Spread Your Bets?

What happened: Epic Systems hosted its annual Users Group Meeting (UGM) last month in Verona, Wisconsin, drawing approximately 7,000 in-person attendees and 37,000 virtual participants. The conference, themed "Storytime," showcased Epic's continued push into AI integration and global expansion.

Key announcements included:

• More than 100 generative AI projects in development or release

• AI-powered drafting of 1 million+ MyChart messages monthly

• AI charting adoption by 180+ organizations

• Introduction of a conversational AI assistant for MyChart

• Expansion of the Cosmos research platform to 270 million deidentified patient records

• Launch of the "Look-Alikes" tool for case-based clinical decision support

• Increased focus on nursing workflows and payer platforms

Why it matters: While the headlines from this year’s event might suggest that Epic announced big leaps forward in AI, the reality is more nuanced.

• Epic mostly highlighted incremental progress, not disruptive innovation:

  Last year, Epic made a big splash with its AI announcements at UGM. This year, by contrast, they’re primarily iterating on existing solutions and encouraging adoption. This suggests it may be time for health systems to invest in leveraging Epic's existing capabilities, rather than waiting for the next big innovation.

• Epic wants to help you integrate AI solutions:

  Epic is investing more in on-site work and integration support, launching a new level-up program earlier this year and expanding its consultative services. This underscores that, even though Epic’s AI tools may be “built in” to your EHR, they still can take plenty of work to implement, requiring significant hands-on effort from both Epic and health systems.

• Epic’s solutions are far-reaching, but they won’t always offer industry-leading performance:

  We’ve heard from executives using Epic’s solutions that its approach of trying to do “everything” AI-related may come at the cost of being industry-leading in any single area. As several health care executives put it, Epic's 'horizontal' nature may ultimately be stifling innovation when it comes to adopting AI solutions. This creates a strategic dilemma: Should you leverage Epic's ecosystem for seamless integration, or should you pursue custom solutions for competitive advantage?

• Data privacy concerns could hinder some of Epic’s big bets

  : While Epic is pushing its Health Grid concept and expanding Cosmos, some systems remain skeptical about patient data sharing. With Epic pushing to have all data interactions backed on their TEFCA platform, concerns about general interoperability and the feasibility of their Health Grid concept remain front-of-mind to many executives. Meanwhile, the success of tools like "Look-Alikes" depends on widespread data integration, which may be hindered by privacy concerns and potential regulatory hurdles.

• Epic’s AI voice assistant demo was impressive … but perhaps not realistic

  : Despite the hype around AI, the executives we spoke to about Epic's new voice assistant had lukewarm feedback. The demonstration assumed smooth integration of multiple complex systems, which may not reflect reality in many health systems.

• For now, you might want to focus on your pain points, not shiny objects:

  Epic's increased attention to nursing workflows and payer platforms suggests a shift towards addressing concrete operational challenges rather than just showcasing futuristic technology. Health systems should align their Epic implementation strategies with their most pressing operational pain points.

Questions to consider:

1. How can your organization better leverage Epic's existing AI capabilities rather than waiting for future innovations?

2. What specialized needs in your health system might require solutions beyond Epic's offerings?

3. How are you balancing the potential benefits of data sharing through platforms like Cosmos with privacy concerns and regulatory compliance?

Augmedix’s Gamble in AI Documentation: ‘Hyper Specialty-Specific Models’

We recently hosted an off-the-record roundtable with Augmedix executives to discuss their AI-powered clinical documentation technology. Here's an inside look at the key takeaways from this confidential discussion.

About Augmedix: Augmedix is a San Francisco-based company that offers a suite of AI-powered documentation solutions and is a leader in ambient technology. The company recently announced a merger with Commure, a health tech company majority-owned by General Catalyst, positioning itself for accelerated growth in the competitive clinical documentation space.

Augmedix executives framed the merger with Commure as a strategic move to accelerate growth and innovation, aiming to double the footprint of large enterprise health systems served. The merger, they said, will bring complementary capabilities, including Commure's revenue cycle management expertise and patient engagement platform.

Augmedix's pitch to health systems: Augmedix executives highlighted several features they believe set their solution apart:

• Flexible implementation options:

  Augmedix offers a three-tiered approach: Augmedix Go (full AI tool), Augmedix Assist (hybrid of ambient AI and human support), and Augmedix Live (full service with dedicated specialists). This allows health systems to mix and match AI and human support, even within a single visit.

• Focus on complex care settings:

  Augmedix argues it is "boldly tackling" highly complex areas and was the first company to deploy solutions for Emergency Departments, Oncology, and Hospitalists.

• CareCues:

  A new feature that can trigger alerts based on certain patient criteria within documents and produce recommendations/action steps. For example, it can alert physicians about the likelihood of aortic dissection when a patient describes migrating chest pain.

• Integration with revenue cycle management:

  Augmedix is developing integrations with RCM systems, potentially linking documentation directly to financial performance.

• Specialty-specific models:

  Augmedix emphasizes the use of "hyper specialty-specific models" rather than general-purpose AI, in order to meet clinicians and health systems where they are.

What your health system peers said:

Health systems are curious about how Augmedix will change ED workflows. A health system CMO asked for granular explanations of how Augmedix's full suite of solutions would impact typical patient encounters in the Emergency Department. He raised concerns about reconciling new AI-driven workflows with legacy clinical decision support tools. While Augmedix highlighted their blending capabilities within existing EMR’s and flagging systems, executives were keen to understand the practical implications of implementation.

There's intrigue, but also hesitation, about the application of ambient technology beyond documentation. Health system executives showed interest in Augmedix's vision for "ambient everywhere," including applications for watches, smart rooms, and nurse safety badges. However, they questioned the practicality and timeline for these advancements. Augmedix acknowledged that many of these applications are still in development, but argued that health systems should be planning for a future where ambient technology is ubiquitous.

Health systems had questions about what the Commure merger will mean in practice. One health system CIO described past experiences of Commure being "scattered" and asked how Augmedix intended to align its offerings within Commure. Augmedix CEO Ian Shakil emphasized recent positive changes at Commure and highlighted the importance of "intent, people, and culture fit" in this merger.

AI Strategy Quick Hits

Noteworthy moves from peers to implement AI technologies

Ambient listening tools, or “AI-powered medical scribing” tools, continue to gain popularity, with health systems like Northwestern Medicine and Kaiser Permanente now partnering with some of the biggest names in the industry.

Since our previous clinical documentation deep-dive webinar in March, this landscape has continued to rapidly expand and evolve. As the industry transforms, how can you ensure you’re making the right choices for your organization? Don’t miss our upcoming webinar on September 11^th to learn about the present and future of AI-powered clinical documentation.

More clinical documentation news:

• Ambience Healthcare launches AI-powered medical scribing technology at John Muir Health

• Southeast Permanente Medical Group implements AI-driven ambient listening tool to reduce physician burnout

• WellPower collaborates with Iliff Innovation Lab to streamline clinical documentation processes

Health systems are strategizing to stay ahead of the curve with it comes to AI investments:

• Why Corewell is taking a 'fast drafter' approach to AI

• Atlantic Health adopts an ‘AI-first’ strategy

• Health systems are looking for ‘quick, early wins’ in RCM

• 3 Health systems discuss when to buy vs build AI solutions

New partnerships are advancing the development of new AI solutions:

• 5 projects are being developed by Children’s National Hospital and Virginia Tech

• Microsoft partners with health systems to streamline nursing workflow

Other strategy quick hits:

• Cerebral reduces chart audit time using AI

• How intelligent document processing automates content-intensive processes

• Press Ganey adds patient surveys to Epic EHR

• Qventus launches AI operational assistants

Emerging Use Cases

New capabilities that indicate AI’s potential

• Mayo Clinic is developing over 200 algorithms to identify specific patient cohorts

• MUSC introduces AI assistant ‘Emily’ to patient access center

• How Memorial Healthcare System is utilizing AI

• AI-powered RPM may improve care for chronic conditions while saving costs

• 10 CIOs share their most unique tech

Cautionary Tales

The risky side of AI implementation

• Cost concerns grow with increases in health system AI adoption

• Why AI testing in healthcare is ‘a mess’

• Which areas of healthcare should be AI ‘no-fly zones’?

• A real-world example of AI ‘drift’ leading to worse results over time

Market moves

A round-up of AI company announcements and stories

• Mount Sinai Health System spin-off, PreciseDX, raises $20.7M in Series B funding

• Investors increasingly favoring generative AI healthcare startups

• Kaiser Permanente and others have invested $10.5M in Clarium

• Optum Ventures participates in a $180M funding for Caresyntax

Policy Updates

Understanding the evolving AI regulatory and legislative landscape

• HHS proposes new regulations for standardizing health IT procurement

• HITRUST unveils new tool for AI risk management

The News in Numbers

An interesting data point that caught our eye

75%

of health system executives believe their organizations are not yet able to capitalize on AI investments due to poor planning or resource allocation. 90% of these executives still believe that AI is a top priority for their organizations.

Expert Insights

For further reading, articles, videos, and podcasts that we found insightful

AI in Healthcare, “C-suite surveyors: AI ‘continues to excite healthcare leaders’” Healthcare IT News, “CIOs and IT leaders must be bold to gain advantage with genAI” STAT, “Can AI help ease medicine’s empathy problem?”

Upcoming AI Catalyst Events

• September 11th: AI Clinical Documentation: What's Now – and What's Next?

  • Discover how early-adopter health systems are transforming clinical documentation with AI and gain actionable insights from top executives and real-world case studies to ensure you’re making the right choices for your organization. We’ll be joined by executives Doug Gentile, SVP, Information Technology, UVM; Lance Owens, CMIO, UMH-West; and Todd Richwine, CMIO, THPG-THR. Click to register.

• September 25th: Deep Dive: How AI is Transforming the Revenue Cycle

  • Explore how emerging AI technologies are reshaping revenue cycle management, hear from early adopters about their experiences and lessons learned, and consider the potential impact to your bottom line. Click to register.