Why AI is such a promising tool for eliminating a hospital’s revenue leakage
There’s ample evidence today that AI has a significant role to play in ensuring a hospital’s financial sustainability by improving revenue cycle performance.
Hospitals today continue to experience revenue leakage, defined as the unintentional loss of income due to inefficiencies, missed billing opportunities and underpayments. Despite their significant investment in electronic health records (EHRs) and billing systems, they continue to lose about 3% to 5% of net revenue annually, amounting to tens of billions of dollars across the nation’s healthcare system.a These losses are especially damaging because most hospitals operate on razor-thin margins, with an aggregate operating margin among all U.S. hospitals of 5.2% after the COVID-19 pandemic.b In this context, even improvements in revenue integrity can be the difference between financial sustainability and insolvency.
AI has emerged as a promising response to this structural challenge, particularly as hospitals confront rising labor costs, payer scrutiny and administrative complexity. Industry analysts are increasingly positioning AI as a critical enabler of revenue cycle modernization rather than a stand-alone solution.c While AI cannot fully eliminate revenue losses, evidence suggests it can materially improve revenue capture, reduce denial rates and lower administrative cost-to-collect through automation, prediction and data-driven decision support.d
By combining machine learning, natural language processing (NLP) and predictive analytics, AI tools can help hospitals identify errors, prevent denials and recover lost income in an unprecedented way. (See the sidebar “Where hospitals lose income in the revenue cycle,” below.)
How AI can help solve revenue leakage
Historically, hospitals have relied on manual audits, staff training and incremental billing system upgrades to address revenue leakage. Yet the sheer complexity of payer contracts and regulatory changes and the vast volume of claims processed each year have severely limited the effectiveness of such measures. AI offers a scalable alternative, enabling hospitals to automatically detect, correct and prevent errors throughout the revenue cycle. Hospital CFOs and revenue cycle leaders can increasingly deploy AI across the following three core domains.
1 Capturing charges. AI can use NLP to analyze physician notes, operative reports and diagnostic documentation to identify services that were performed but not billed.
2 Predicting and preventing denials. AI employs machine learning models to forecast prior to submission the likelihood that claims will be denied, giving billing staff the opportunity to correct errors in advance.e In one anecdotal report to the author, a hospital using denial prediction tools experienced a19% reduction in denial rates within six months.
3 Ensuring contract compliance and recovering underpayments. AI-enabled contract compliance tools can analyze large volumes of remittance data to identify systematic underpayments and payer deviations from negotiated rates, providing hospitals with evidence for appeals and contract renegotiations.
Regulatory support for AI
Policy and regulatory shifts are also accelerating the adoption of AI in hospital finance. CMS is currently rolling out its WISeR (Wasteful and Inappropriate Service Reduction) Model aimed at reducing clinically unsupported care.a CMS states that WISeR aims to “Protect American taxpayers by leveraging enhanced technologies, such as artificial intelligence (AI) and machine learning (ML) … to ensure timely and appropriate Medicare payment for select items and services.”
Under the FY 2024 inpatient prospective payment system final rule, social determinants of health such as homelessness or food insecurity now influence payment rates. AI can play a critical role in ensuring accurate documentation of these factors, preventing under-reimbursement. At the same time, hospitals must ensure that AI tools comply with HIPAA, the False Claims Act and Stark Law. This regulatory environment underscores the importance of strong governance, auditability and risk management in AI implementation.
Footnote
a. CMS, “Medicare Program; hospital inpatient prospective payment systems for acute care hospitals and the long-term care hospital prospective payment system and policy changes and fiscal year 2024 rates,” Federal Register, Aug. 28, 2023.
Case examples of effective AI use in preserving hospital revenue
Evidence of the benefits of such AI interventions is mounting.
A wide variety of articles and reports have been published that highlight benefits that have been achieved by applying AI solutions to lower denial rates. The cover story of this issue of hfm is just one example. (See also the sidebar at the end of this article for additional examples of recent studies and reports.)
To these examples, I would add a number of real-world examples that have been shared with me anecdotally by healthcare leaders.
A large academic medical center in California implemented an AI tool with NLP to cross-checkclinical notes against billed charges. Within six months, the system identified $12 million in missed charges, primarily related to ancillary services such as laboratory and imaging. Coding-related denials also fell by 22%.
A Midwest health system with $3 billion in annual revenue deployed a denial prediction AI model integrated with its EHR. By proactively flagging high-risk claims, the system reduced denial rates by 18% and improved first-pass yield from 85% to 92%, translating to $40 million in additional net revenue in a single year.
An integrated delivery network in Texas used AI to analyze payer compliance and uncovered recurring underpayments for orthopedic procedures. Armed with this evidence, the CFO successfully renegotiated payer contracts, yielding an 8% increase in reimbursement worth more than $25 million annually.
Challenges to AI adoption
Despite its promise, AI adoption in revenue cycle management is not without challenges. Data quality remains a primary concern, as poorly structured EHR data can compromise the accuracy of AI models. Seamless integration between AI platforms and existing EHR or billing systems also presents technical hurdles.
Workforce adaptation poses another challenge: Staff roles must shift from manual billing to overseeing and validating AI outputs.
There are also ethical and legal risks to consider. For instance, if AI errors result in overbilling, hospitals could face liability under the False Claims Act.
Moreover, while upfront costs of AI adoption can be significant, it can take time to see meaningful results. Most systems see ROI within 12 to 24 months.
Recommended actions
To maximize the benefits of AI, hospital CFOs should ensure their organizations pursue a deliberate, phased strategy. For this purpose, they can benefit from adopting the following strategic recommendations.
1 Establish baseline leakage measurements. Hospitals should perform denial analyses, missed-charge audits and underpayment reviews, thereby providing benchmarks against which to evaluate AI’s impact.
2 Prioritize quick wins. Hospitals can do so by deploying AI in areas such as eligibility verification, claim scrubbing and denial prediction — domains where measurable results can be achieved within six months.
3 Build governance and compliance into AI adoption from the outset. Hospitals should establish oversight committees with clear audit trails and model-validation protocols to ensure compliance with HIPAA and other regulations.
4 Scale efforts systematically. Hospitals should begin, for example, with front-end denial prevention before moving into advanced applications like NLP-based documentation improvement and AI-assisted coding.
5 Track performance through weekly dashboards. These dashboards should incorporate key metrics such as first-pass yield, denial rates, discharged-not-final-billed accounts, days in accounts receivable and patient self-pay collections.
6 Prepare for workforce transformation. Such preparations should include staff training to transition from manual data entry and billing tasks to AI monitoring, exception handling and financial analysis.
7 Leverage AI insights in payer negotiations. This use of AI is important for reducing systematic underpayment through data-driven contract management.
AI: An essential tool for healthcare’s future
Revenue leakage has long undermined U.S. hospital finances, silently eroding margins in a sector already under immense pressure. While AI cannot fully eliminate leakage, it has proven to be a transformative tool, capable of producing meaningful results in recovering lost revenue, reducing denial rates and cutting administrative costs.
For healthcare CFOs and other executives, the strategic imperative is clear: AI adoption is no longer optional but necessary.
By embedding AI systematically across the revenue cycle, hospitals can shift leakage from an uncontrollable drain into a manageable, shrinking cost center. Beyond improving financial resilience, these gains ultimately support what is necessarily the core mission of every health system: to deliver high-quality, sustainable patient care.
Footnotes
a. Thakur, S., “The hidden billion-dollar problem in U.S. healthcare: Why underpayments matter more than denials — and how AI can fix it,” Medium, Jan. 12, 2026.
b. Levinson, Z., Godwin, J., and Neuman, T., “Hospital margins rebounded in 2023, but rural hospitals and those with high Medicaid shares were struggling more than others,” KFF, Dec. 18, 2024.
c. See, for example, HFMA, “How leveraging artificial intelligence in utilization management can enhance your revenue cycle,” research report sponsored by XSOLIS, July
27, 2023.
d. Russell, S., Suros, F., and Kumar, A., “Exploiting machine learning bias: predicting medical denials,” Proceedings of the 2024 AAAI Spring Symposium Series, May 20, 2024.
e. Hammer, D., “Accelerating RCM with AI breakthroughs,” presentation to HFMA’s Hawaii Chapter, March 22, 2024.
Where hospitals lose income in the revenue cycle
Hospitals can experience revenue leakage in multiple points in the revenue cycle, spanning front-end, mid-cycle and back-end processes.
Front-end problems. Claims often can be denied or payment delayed as a result of errors inpatient registration, insurance eligibility verification and prior authorizations. Nationwide benchmarking data indicate that front-end revenue cycle issues have accounted for as much as 46% of all denied claims.a
Mid-cycle deficiencies. Mid-cycle processes can present vulnerabilities in the form of incomplete clinical documentation, inaccurate coding and missed charges.
Back-end shortcomings. On the back end, revenue leakage can result from claim denials, payer underpayments and weak follow-up practices. Approximately 15% of claims are denied at first submission, and nearly two-thirds of these are never resubmitted.b Moreover, hidden underpayment within payer contracts can erode collections by as much as 11%.c
Footnotes
a. Change Healthcare, The Change Healthcare 2022 Revenue Cycle Denials Index, 2022.
b. American Hospital Association, “Payer denial tactics — How to confront a $20 billion problem,” AHA Center for Health Innovation Market Scan, April 2, 2024; and Poland, L., and Harihara, S., “Claims denials: A step-by-step approach to resolution,” Journal of AHIMA, April 25, 2022.
c. Narasimhan, S., “The hidden cost of underpaid claims and how analytics can plug the gap,” blog, HealthAsyst, Sept. 11, 2025.
AI’s potential for reducing denials: Studies and reports
- Williams, J., “Battle of the bots: As payers use AI to drive denials higher, providers fight back,” hfm, April 2024, updated online Dec. 22, 2025.
- American Hospital Association, “Optimize your hospital’s revenue cycle for efficient, patient-centered operations,” sponsored by R1 RCM, 2024.
- Pennington, R., “Artificial intelligence (AI) and its opportunity in healthcare organizations revenue cycle management (RCM),” Marshall University, Theses, Dissertations and Capstones, 2023.
- Owolabi, T., “Transforming appeal decisions: machine learning triage for hospital admission denials,” JAMIA Open, Feb 25, 2025.
- Rego, R., “AI in denials management and prevention: A strategic imperative,” BerryDunn, June 24, 2025.
