Health systems are eager to adopt generative artificial intelligence solutions for revenue cycle management, seeing the promise of the technology to improve coding and capture more revenue.
Eighty percent of health systems say they’re exploring, piloting or implementing gen AI tools for RCM in 2025, a 38% jump in less than two years, according to a new survey by the Healthcare Financial Management Association (HFMA) and AKASA, a health tech company that provides gen AI solutions for hospital RCM. In 2023, 58% of health systems were merely considering gen AI for the revenue cycle.
About 40% of health systems are ahead of the game, either piloting or implementing gen AI tools in their operations. Another 39% are exploring their options, the latest survey found.
Commissioned by AKASA, the survey fielded responses from 519 chief financial officers and revenue cycle leaders at hospitals and health systems across the U.S. through the HFMA Pulse Survey program in April 2025. AKASA provided Fierce Healthcare with a first look at the survey results.
As health systems are navigating a challenging financial environment, the survey results match the general sentiment from healthcare finance leaders that AI could be a crucial tool to help improve RCM operations, said Jeff Francis, chief financial officer of Methodist Health System in Omaha, Nebraska.
“I was at a conference last month, and the moderator noted that two years ago, at that same conference, the AI panels were about how scary it was going to be, and this year, it was the excitement and the opportunities around it. It really has changed in two years; I’m definitely seeing it on the revenue cycle side,” Francis told Fierce Healthcare.
The healthcare industry is ready to implement AI for RCM but is still grappling with deeply rooted operational barriers, the survey found.
Many health systems remain in the early stages of adoption, citing cost and budget constraints as the largest obstacles as providers bump into the realities of implementing new AI technologies at scale.
One in 5 organizations (20%) have not yet begun their journey with gen AI for revenue cycle management. Among smaller health systems, those with revenues between $500 million and $1 billion, about 20% are piloting and implementing, compared to more than half of larger health systems (64%), the survey found.
Midsized health systems—those with annual NPR between roughly $500 million and $1 billion—are among the most cautious adopters, often constrained by competing priorities and limited IT capacity, according to the survey.
“AI isn’t the barrier. Resources are,” one health system executive said in the survey. “Teams need time, education, and support to make adoption sustainable.”
Health systems cited integration with existing systems as a major barrier as they roll out AI RCM solutions as well as security and privacy concerns. Health systems executives also reported lack of a clear return on investment.
To address many of these issues, health tech vendors need to minimize the heavy lifting for health systems to adopt AI solutions, said Malinka Walaliyadde, co-founder and CEO of AKASA.
“Use standards-based integrations. All of these health systems are typically using some type of electronic health record that has these accessible APIs (application programming interfaces). As an industry, we invested heavily into making those APIs available. If you minimize the lift on their side from an IT side, especially for smaller health systems, these things become accessible,” Walaliyadde told Fierce Healthcare. “As a technology company, we and others in our domain should invest in using those and then using AI to do the heavy lifting on our side versus asking health systems to do the heavy lifting on their side.”
Gen AI solutions are proving their value by strengthening the accuracy and completeness of documentation and coding that directly influence quality scores, compliance and revenue integrity.
“The opportunity to improve accuracy in documentation is clear, but it requires investment, trust and a roadmap that integrates technology into existing workflows responsibly,” Walaliyadde said.
The cost of implementing AI solutions continues to be a key concern among finance leaders, Francis said. Inflation, changing regulations, payer mix shifts and workforce shortages continue to squeeze already tight margins for health systems.
“It does take a little bit of investment and maybe a little bit of time before you start seeing the results. A lot of times, if you’re in a smaller health system, you just may not have that luxury to do that. It’s being able to get over that initial financial hurdle to start receiving that benefit,” he said.
Interest in gen AI solutions for RCM is growing as health systems continue to face pain points in midcycle RCM operations with documentation and coding. More than a quarter of healthcare leaders identified incomplete documentation as their most significant challenge, surpassing both staffing and technology limitations.
Challenges with documentation accuracy and completeness can have real financial implications. The majority of the finance leaders surveyed (89%) said missed or inaccurate codes impact revenue, and half (51%) described the impact as significant or very significant.
Both small and large health systems report meaningful revenue at risk due to these errors. On average, survey respondents estimated that 8.49% of total revenue is at risk due to documentation or coding issues.
“That’s huge,” Francis said. “Think of 8% of revenue on a $1 billion organization. So you’re talking impacting $80 million and on tight margins, that’s really cutting in to the ability for health systems to reinvest into infrastructure, being able to grow or innovate because that is really keeping money going into research or capital investment.”
When asked where gen AI could make the greatest difference, finance leaders cited opportunities to identify missed reimbursement opportunities (nearly 60%), uncover gaps in clinical documentation (57%) and identify missed quality indicators (33%).
Methodist Health System has been working with AKASA since 2019 to speed up claims resolution. The technology automated revenue cycle tasks, allowing the health system to shift resources from insurance follow-up activities to other revenue cycle functions. Through the use of the technology, 71% of accounts were removed from staff queues, doing the status work of nearly 14 full-time employees, according to a case study. The AI solution resolved claims for 56,118 accounts in eight months, saving 5,559 hours of work during that period, the company said.
“We’ve been able to free up eight FTEs’ worth of time so they’re not doing value-add work of looking at claim status on a payer’s web portal, but actually diving in to work on the more difficult claims, and we’ve seen an improved yield because of it,” Francis told Fierce Healthcare.
The health system is now looking at opportunities to automate the pre-authorization process, he noted. “We already use it on coding review. What we’ve done in the past is we would have a claim come back or it would be denied, or we would think that we were underpaid, and we would then send it out to a third party. That’s months down the road. We’ve now been able to implement a tool that it’s done pre-bill so before it even goes out. We’re able to get a more accurate claim and get it right the first time to get those dollars quicker to the organization,” he said.
AKASA developed its technology by combining revenue cycle knowledge with custom large language models (LLMs) trained on a health system’s own clinical and financial data. The aim is to use LLMs to “help capture the patient story as comprehensively as possible for health systems, and then tell that story to payers,” Walaliyadde said.
More accurate documentation and coding enable health systems to get credit for 100% of the services they provide, producing better financial and quality outcomes while delivering more precise care, according to AKASA executives.
“Our goal is to make sure that the patient story is documented as accurately as possible, and then code it as accurately as possible. If we do those things correctly, everything else will work itself out. Because, in reality, when we do these things, we will both make sure things that should be captured are captured, but also if things are captured that shouldn’t be, we will flag those as well,” Walaliyadde said. “The North Star is accuracy. AI can do a substantially better job. When you do that, all of these other things fall into place.”
AKASA says its AI-native solutions are quickly gaining traction among health systems; the company works with many large health systems including Cleveland Clinic and Duke University Health System. Cleveland Clinic has expanded its mid-revenue-cycle-focused partnership with the company, rolling out its AI-powered clinical documentation integrity tool across its U.S. locations.
Publisher: Source link
