Healthcare leaders optimistic that automation and AI will improve revenue integrity

As the cost of clinical documentation and coding discrepancies rises, many are turning to automation and AI with human oversight to solve the problem. 

More than a third (37%) of healthcare organizations say clinical documentation and coding discrepancies have a negative financial impact on their organization of at least $1 million. Moreover, 12% estimate the impact at more than $5 million, according to new research conducted by HFMA and analyzed by Solventum.  

However, survey data also indicates an overall sense of hope among leaders that automation and AI will strengthen revenue cycle accuracy and efficiency and boost revenue integrity. Ninety percent of survey participants believe that with human oversight, these technologies will be “moderately” or “extremely” effective at improving financial performance. The survey, conducted between May 16 and June 2, 2025, includes responses from 272 participants who are healthcare directors, presidents, CFOs, managers and other members of HFMA. Among the findings:  

• About half of organizations have already implemented AI-driven solutions in mid-revenue cycle workflows (e.g., coding, denials management, clinical documentation improvement [CDI]).  
• Two out of five organizations always address denials retrospectively.  
• One out of three organizations are testing, evaluating or in the process of implementing coding automation. The same is true for denials management automation (32%) and insurance discovery/prior authorization automation (38%).  
• Only 9% of respondents say they are “very confident” their organization captures all the revenue to which it is entitled.   
• The top three barriers to adopting AI-driven revenue cycle solutions are cost (57%), lack of IT resources (55%) and unknown ROI (41%).  
• While two-thirds of respondents say medical necessity or prior authorization are the most common reasons for denials, 60% need more insight into the root causes of denials to improve revenue integrity efforts.  

“If more than a third of organizations are seeing a seven-figure financial impact from documentation and coding discrepancies — and only 9% are confident they’re capturing all entitled revenue — it’s clear the status quo isn’t working,” said Jason Burke, global vice president of revenue cycle at Solventum. “Additionally, payor requirements and denial trends are a moving target, so leveraging AI and machine learning to understand changes in as close to real time as possible has now become a necessity to maintain financial viability.”   

It’s an area where laying the right foundation for AI and automation in revenue cycle — one that includes prospective workflows, consolidated tools and a clear plan for promoting data integrity using technology and human oversight — is critical to success.  

Embracing prospective workflows  

While retrospective workflows, particularly those pertaining to denials management, are still common, Burke said the danger of backward-looking processes is that they often ignore the broader operational context, making them less effective and more time-consuming. “For example, retrospective approaches to denials management treat denials as isolated financial transactions instead of signals that process improvement is necessary,” he said.  

Instead, organizations wanting to promote prospective workflows should consider the following steps:  

1. Dig into the root causes of denials. The ability to understand the root causes of denials represents a significant knowledge gap for most organizations. “It’s not that the insights aren’t there; it’s that they’re hidden in silos or only visible after revenue is already lost,” Burke said. To make meaningful progress, he said organizations must:  

• Apply root cause analysis at the workflow level.  
• Centralize visibility into the denial lifecycle.  
• Make feedback loops operational.  
• Standardize denial reason mapping across IT systems, including coding, CDI, billing, denials management and communication.  
• Use AI to surface predictive risk factors.  

2. Ensure real-time visibility into documentation quality. “Without seeing documentation completeness or clarity before coding, you’re stuck fixing problems after they occur,” Burke said. 

For example, organizations can integrate medical necessity content consistently and reliably during scheduling, registration and claim processing, reducing guesswork around what a payor may or may not accept.  

3. Integrate clinical and revenue cycle teams in addressing revenue integrity breakdowns. “Retrospective models often leave CDI, coding, utilization review and billing teams in silos, each solving their own piece too late,” Burke said. Critical to this effort: establishing process ownership and accountability. 

4. Leverage predictive analytics and automation during the inpatient encounter. “Many organizations aren’t leveraging AI to flag high-risk encounters or recommend fixes while the patient is still in-house,” Burke said. 

5. Provide physician education that’s rooted in denials data. “Retrospective models often fail to close the loop with clinicians about their documentation, thus missing an opportunity for behavior change,” he added.  

Consolidating revenue cycle solutions  

While many organizations continue to use multiple standalone revenue cycle management (RCM) solutions, Burke says consolidated revenue cycle systems provide the structured foundation and end-to-end visibility organizations need to drive enterprisewide revenue integrity.  

In outsourcing revenue cycle services, “Organizations should consider partnering with an RCM vendor that not only accurately and compliantly codes their services, but also one that alerts them to potential denials and payment reductions or delays,” he said. “Speeding the process of coding only to result in post-denial rework is not a win. It is simply a shift of the workload in the most inefficient way.”  

Working with a single RCM vendor that can solve more than one business problem in the RCM space means fewer interfaces, fewer systems to maintain and less training to provide. However, that vendor also must be able to provide solutions that keep teams aligned and reduce rework using consistent, defensible and compliant intelligence.  

“Everyone, from clinicians to coders to auditors and others, must feel confident that a complete and accurate record has been generated, coded and billed for in a compliant and efficient way,” Burke said.   

How can healthcare leaders select the right partner? Burke recommends asking three questions:  

• Does the solution provide a single source of truth for documentation, coding and denials data or must teams reconcile information across multiple systems?   
• Are compliance updates applied consistently across all modules (e.g., coding, CDI, denials) within the platform?   
• What reporting is available so leaders can measure ROI?  

Addressing common barriers  

As the top barrier to adopting AI-driven revenue cycle solutions, cost is something healthcare leaders must approach carefully.  

“With autonomous coding in particular, cost may vary between organizations based on case volume, specialties covered, integration requirements and whether the organization is replacing outsourced services or augmenting internal teams. However, the ROI can be substantial,” Burke said.  

He offers the following strategies to address cost- and ROI-related barriers specifically associated with autonomous coding:  

• Use a phased rollout. Begin with targeted service lines to manage upfront costs while delivering early wins that fund broader adoption. “This approach not only contains expenses, but also builds a sustainable case for long-term savings and efficiency gains,” he said.  
• Compare AI-driven performance against previous coder benchmarks. These include coder productivity, accuracy, turnaround time, discharge-not-final-billed and discharge-not-final-coded. “Many organizations underestimate the power of benchmarks to build confidence in ROI. Comparing baseline metrics can highlight tangible improvement opportunities,” Burke said. In addition, organizations should see a reduction in reliance on outsourced coding and the ability to redeploy staff on complex cases.  

Privacy and security concerns are also a top barrier to AI and automation adoption. In addition to verifying that a vendor complies with standards such as SOC-2, GovRAMP or FedRAMP, Burke recommends asking the following questions:  

• How frequently does the vendor conduct penetration tests, vulnerability scans and third-party audits?  
• What is the vendor’s process for breach detection, notification and incident response?  
• How does the vendor limit and monitor access to sensitive data within the platform?  

Survey findings show why these questions matter: 61% of organizations cited risk of protected health information breaches as their top security concern when adopting healthcare technology.   

“It’s not just about regulatory compliance,” Burke said. “Patients trust providers to safeguard their information, and any breach can have long-term reputational and financial consequences.”  

In addition, lack of IT resources is a common barrier to AI and automation. IT staff often are already stretched thin managing EHR upgrades, infrastructure and cybersecurity, leaving little bandwidth to take on new revenue cycle projects. Cloud-based autonomous coding solutions that integrate directly with existing EHR and billing platforms can significantly reduce this burden.   

Burke added that organizations should look for vendors that not only deliver the technology, but also provide managed integration, regular updates and hands-on implementation support.  

“The right partner makes IT enablement lighter, not heavier,” he said.  

Focusing on data integrity using human oversight  

“There’s a lot of great technology available on the market to automate low-hanging fruit; however, we still need auditors and humans to validate content, identify root causes and provide content to train the technology,” Burke said. “For example, with autonomous coding, we see the role of the medical coder evolving to become auditors and exception coders. It’s a role where they would review the work of the ‘AI coder’ through a built-in QA [quality assurance] process and use their expertise to focus on complex cases that don’t qualify for automation.”  

Similarly, just as coders are evolving into auditors and exception coders with autonomous coding, other mid-revenue cycle roles are also shifting as AI takes on repetitive work. For example, a CDI specialists spend less time flagging routine documentation gaps and more time on complex cases where provider education, clinical nuance and judgment are essential. Their role becomes one of guiding providers, validating AI recommendations and helping ensure that documentation reflects the true clinical picture.  

“Across coding, CDI and denials, AI doesn’t replace expertise,” Burke said. “It shifts the focus from repetitive review to higher-value analysis, oversight and education, ultimately strengthening both compliance and financial performance.”  

Looking ahead 

As healthcare leaders look ahead to prioritize AI and automation efforts, the most important question to consider is: Where are manual efforts high and variable — and where do they directly affect revenue?  

“Automation makes the biggest impact when it replaces repetitive, error-prone tasks that tie up skilled staff or slow down the revenue cycle,” Burke said. “Ultimately, the goal isn’t automation for its own sake; it’s to create smarter, more resilient workflows that reduce revenue risk and help people focus on the work only they can do.” 

About Solventum

Solventum, formerly 3M Health Information Systems, is built on a legacy of innovation and commitment to creating breakthrough solutions for our customers’ biggest challenges. By connecting clinician documentation and revenue cycle workflows we unlock efficiency, improve accuracy, and eliminate rework to compliantly improve financial and quality outcomes. We deliver expert guided technology to transform clinician documentation with ambient and speech understanding technology. We connect that directly to AI-driven revenue cycle solutions including proactive clinical documentation integrity, automated coding, and denials prevention. Our proven clinical methodologies help improve quality, safety and value across all care settings. From capture to code and beyond, we never stop solving for you. See how at Solventum.com/HIS

_{This published piece is provided solely for informational purposes. HFMA does not endorse the published material or warrant or guarantee its accuracy. The statements and opinions by participants are those of the participants and not those of HFMA. References to commercial manufacturers, vendors, products, or services that may appear do not constitute endorsements by HFMA.}