It’s Time for Meaningful Action on Medical AI

—
Congress and HHS must align payment models to support AI integration

by

Richard Heller, MD, MBA, and Nina Kottler, MD, MS

April 19, 2025 • 5 min read

It is the policy of the United States to sustain and enhance America’s global AI dominance in order to promote human flourishing, economic competitiveness, and national security. — Executive Order on “Removing Barriers to American Leadership in Artificial Intelligence,” The White House, January 23, 2025

One of the most talked about disruptors in healthcare today is artificial intelligence (AI) and the power of technology to fundamentally change the practice of care delivery. AI was brought up by Mehmet Oz, MD, MBA, the physician, talk show host, and new CMS administrator, in his Senate confirmation hearing, as a way to address many of the challenges facing our healthcare delivery system. While much has been said about the promise of AI, details of how it should be implemented and the role of the government in promoting adoption are less granular.

To be clear, if we as a nation fail to invest in the meaningful use of AI, we do so at our own peril. Congress should work with HHS to promote use of AI technologies that improve the value of care by aligning payment models to support their integration.

While AI has the potential to transform numerous industries, its impact on healthcare can be particularly profound. From assisting physicians in identifying critical findings on medical images to enhancing doctor-patient communication, AI is poised to revolutionize the way medicine is practiced.

A recent physician survey on AI revealed that reducing administrative burden is seen as the greatest opportunity for AI among more than half of physicians (57%). Indeed, the most common AI applications in healthcare involve tasks such as automating billing documentation and drafting medical notes. These tools are expected to “pay for themselves,” primarily through efficiency gains.

But what about AI tools that improve patient care without increasing efficiency or those that even make physicians less efficient? Who should cover the cost of these innovations? And without reimbursement, will these tools be widely adopted?

The federal government has two critical roles to play in the use of AI in medicine: ensuring the safety and efficacy of tools that impact patients, and establishing fair reimbursement policies to support their adoption. The first responsibility lies with the FDA, which has been diligently working to keep pace with rapidly evolving technologies. The FDA has sought input on its draft proposals, convened public advisory committee sessions to engage stakeholders, and conducted site visits to gain a deeper understanding of real-world implementation and functionality. For example, in 2024, more than a dozen members of the FDA’s Center for Devices and Radiological Health (CDRH) visited our practice as part of its Experiential Learning Program. However, even with optimized FDA oversight, the full benefit of these tools won’t be realized without adequate reimbursement mechanisms to facilitate their integration into clinical practice.

Many AI companies are selling directly to hospitals and medical practices, with a value proposition based on a financial return on investment. This business model attempts to capture (for the AI vendor) a portion of the additional revenue generated by the hospital or practice through greater clinical volume (e.g., increasing the number of daily clinic visits) or a fraction of the resultant cost savings (e.g., reduced staffing needs).

From a federal perspective, while there is interest in supporting AI, CMS has granted payments for a limited number of tools that meet strict criteria. Unfortunately, CMS’s payment systems are not optimally designed to support innovation. A recently proposed bill would be useful, however it is limited to technical payments for AI tools utilized in the hospital-based outpatient setting. With over 75% of the AI tools authorized for marketing by the FDA focused on radiology, and given the fact that we’ve had nearly a decade of experience with AI in our own practice, we as radiologists are ideally positioned to identify barriers to adoption and offer suggestions for improvement.

While quality-focused AI has the potential to transform the care delivered to patients, the high costs of acquisition and implementation, coupled with the fact that efficiency gains, if any, may be limited and delayed, make adoption difficult — especially in a challenging economic environment. For context, the payment rate for Medicare providers in 2025 is the lowest it has been in over 30 years. When paired with aggressive tactics from some corporate health insurance companies, many medical practices lack the financial resources to invest in AI, even if it benefits patients. Some organizations have resorted to charging patients directly for the use of AI on their exams. Other institutions, fearful that “up-selling” patients could exacerbate care inequities, pay for it out of their pocket but acknowledge that this approach is unsustainable.

As Congress and HHS consider more comprehensive approaches to achieving the goal of the executive order, we believe that programs should adhere to the following principles:

1. Support the adoption of AI tools that improve patient care by: (a) Elevating quality of care; and/or (b) Increasing access to care; and/or (c) Reducing costs of care.
2. Ensure that improvements are quantifiable and meaningful.
3. Maintain fiscal responsibility.

Earlier this century, our healthcare system faced a similar challenge with the transition to electronic health records (EHRs). Like AI, adoption of EHRs is associated with high upfront costs but long-term benefits to the system. Therefore, an EHR incentive program, known as Meaningful Use (MU), was developed. The MU program utilized early incentives that declined over time (eventually penalties replaced incentives). Using a similar approach of front-loaded, declining incentives, an AI-focused program could help accelerate adoption and implementation, ensuring interoperability, and promoting responsible use. Such an approach could complement the suggestions from health economists.

While the U.S. healthcare system is far from perfect, one of its shining achievements has been its ability to offer patients the latest technological advancements in diagnosis and treatment. The executive order on AI is a strong statement, but translating it into meaningful use requires establishment of a robust payment pathway for AI technologies. Until the federal government takes decisive action, American physicians — and the patients we serve — may be left waiting, unable to fully realize the benefits of AI-driven healthcare innovations.

Richard Heller, MD, MBA, is senior vice president of health policy at Radiology Partners. Nina Kottler, MD, MS, is a radiologist and associate chief medical officer of Clinical AI at Radiology Partners.