Trends 2025: AI in healthcare progressing despite reimbursement hurdles

Photo: Westend61/Getty Images

Artificial intelligence is poised to become a major force in healthcare. The technology is developing at breakneck speed, and both payers and providers are examining the potential use cases, with one survey from November showing 73% of organizations plan on growing their financial commitments to the technology. But because AI is so new, there are still some knots left to be untangled.

One of the biggest knots is reimbursement – specifically, the reimbursement of new AI-driven devices and processes. Reimbursement is paramount to improving access to advanced tools, but even for FDA-cleared devices, the path to reimbursement for these technologies is unclear and slow-moving. 

That, said Dr. Brit Berry-Pusey, can hinder the adoption of the latest AI devices.

Berry-Pusey is the chief operating officer for AI cancer-mapping startup Avenda Health. While Avenda has been able to secure Medicare reimbursement for its products, it hasn't been an easy journey, as it was only the fifth AI startup to achieve this milestone.

Currently, the way the approval process for new devices is set up, it can take a full seven years between FDA clearance and the time reimbursement is finally established.

"The FDA has learned a lot about how to regulate products, and the government has learned how to pay for AI products," said Berry-Pusey. "Unfortunately, the way reimbursement is set up in the U.S., it disincentivizes new technologies. If you're really pushing the boundaries and creating something novel, it means you have to start from scratch from a reimbursement perspective."

It can be a lengthy process. A company begins by first submitting an application, with applications only accepted during four windows per year. It then heads to a review, during which a panel votes yes or no on the product; after a yes vote, it takes about a year for the code to take effect. Then the company goes through the payment process and all of the other information that needs to be submitted to the Centers for Medicare and Medicaid Services. Without a clear deadline, the company won't know exactly when they can expect a response from CMS.

This hypothetical AI company can get a temporary code for their product, but there's a high burden of evidence that the technology is working, even if the FDA has already given the green light. 

Providing additional evidence can be challenging for startups and smaller players, said Berry-Pusey.

"It's a lot of nuance, and a lot of complications, and it's not benefitting people," she said. "It can be challenging. It's not easy for small companies."

Berry-Pusey hopes the situation improves, both because of AI's potential to help patients, and because the industry is rapidly heading toward adoption of the technology – and increasing its investments.

POISED TO GROW

A Define Ventures survey from last month detailed the extent to which this is true, finding that three-quarters of health system leaders and insurance executives plan on investing more in AI in the future.

Also, 73% of organizations have established governance structures, which can align AI incentives with organizational values. The primary focus areas for these governance committees include identifying and prioritizing use cases (91%), establishing ethics and safety guidelines (87%), and setting data policies (84%).

Payers and providers both view AI as a transformative force in healthcare, with 54% expressing confidence in its potential to reshape patient and clinician experiences in the next few years.

Thirty-three percent believe AI will decrease healthcare costs, 9% believe it will improve quality of care, and 4% believe it will improve access. That helps explain why 76% are establishing AI pilot programs that focus on small-scale projects to validate AI's impact before committing to wider adoption. Another 71% are in the process of identifying specific use cases, which indicates that, while AI shows potential, leaders are still being deliberate about its implementation.

At the same time, three-quarters of U.S. healthcare providers and payers increased their IT spending this past year, with artificial intelligence, cybersecurity and IT infrastructure among the chief areas of investment, according to a study by Bain & Company and KLAS Research. 

Adoption of AI is gaining traction, with 15% of providers and 25% of payers reporting an established AI strategy in 2024, the report said.

One of the use cases for AI in healthcare that has started to emerge is improving revenue cycle performance. A September Guidehouse analysis showed that nearly half of healthcare leaders reported a 93% or less net collection yield, representing a significant opportunity to improve performance. Executives said their highest priority for revenue cycle investment in the next 12 months is technology, such as AI, automation and machine learning, though they were concerned about cybersecurity and the potential for ransomware attacks.

Statistics provided by TruBridge, a consultant and IT services provider, at the HFMA annual conference in Nashville in 2023 found there's $9.8 billion in potential savings through AI-powered automation in the revenue cycle. Nine percent of all claims are rejected in error or from prior authorization denials; 23.9% of denials are due to eligibility issues.

Seventy-five percent of hospitals are working on an AI strategy to tackle revenue cycle challenges, said Patrick Murphy, general manager of TruBridge and the former CFO of a health system in Alabama.

In March, the Congressional Budget Office addressed AI and machine learning in healthcare, and determined that the evidence on the usefulness of the technology is mixed, particularly when it comes to costs. 

The practical application of these technologies is still inconsistent at this nascent phase, CBO determined. It said it will need to see more empirical evidence before determining the overall effect on things like healthcare spending.

Even so, adoption of AI is gaining traction, with 15% of providers and 25% of payers reporting an established AI strategy in 2024, found Bain & Company and KLAS Research in September.

Major players in the healthcare industry are getting in on the act. Just this year, Cleveland Clinic and the Novo Nordisk Foundation launched the Cleveland Clinic-Denmark: Quantum-AI Biomedical Frontiers Fellowship Program, which is dedicated to integrating quantum technologies and AI into biomedical research and patient care. 

Meanwhile, Minnesota-based Mayo Clinic is launching a new $10 million artificial intelligence education program that will train staff and medical professionals to deploy the best AI technology, ethically, for patients, while Mount Sinai Health System in New York is opening the Hamilton and Amabel James Center for Artificial Intelligence and Human Health, which is dedicated to the research and development of AI tools and technologies.

Clearly, the healthcare industry is leaning into the technology. But Berry-Pusey hopes it will soon be easier for startups to innovate and muscle into the space.

THE FUTURE

There are a few different proposals to improve the reimbursement scenario for AI companies, said Berry-Pusey. One that was floated several years ago, but did not pass, would have given special designation to breakthrough devices. It would have granted temporary reimbursement for up to three years to help the products accrue additional data, and to go through the standard processes that are required. 

These, said Berry-Pusey, would have been technologies that were already FDA-approved and shown to be safe.

Another similar pathway has been proposed, but it would be limited to five technologies per year and wouldn't be of much help to startups, said Berry-Pusey. So, the reimbursement picture for AI startups is stuck in a strange state of limbo.

"I do hear rumors that Congress is looking at something similar to the original proposal," said Berry-Pusey. "Anything we can do to make sure patients have access to technology would be so beneficial, because unfortunately the challenges of reimbursement contribute to inequities. The only patients who have access to them are the ones who can pay out-of-pocket. Even patients who have great insurance or are covered by Medicare or private care, they don't have access to the technology unless they have access to cash."

According to Berry-Pusey, the challenges or reimbursement will hinder some of the innovation in this space. There's still innovation, but it may not be as fast or enticing as it could be if there were a clearer path, she said.

Still, helping others is a big motivator, and there are still companies such as hers that are willing to endure the challenges.

"It's a really, really exciting time," said Berry-Pusey. "Hopefully more and more people who have the power and ability to change some of the hurdles that we're currently facing will see the value, and will improve the process and improve the system."

Jeff Lagasse is editor of Healthcare Finance News.
Email: jlagasse@himss.org
Healthcare Finance News is a HIMSS Media publication.