Artificial intelligence (AI) continues to dominate discussions in the health care industry, not just about innovation and implementation, but also about how it should be regulated. As announced by the Centers for Medicare and Medicaid Services (CMS), on April 4 CMS finalized the “Contract Year 2026 Policy and Technical Changes to the Medicare Advantage Program, Medicare Prescription Drug Benefit Program, Medicare Cost Plan Program, and Programs of All-Inclusive Care for the Elderly.” While predominately a technical rule for Medicare Advantage and the Part D program, it had a notable passage on what wasn’t finalized (emphasis added):

We also do not intend to finalize the following provisions from the proposed rule: Enhancing Health Equity Analyses: Annual Health Equity Analysis of Utilization Management Policies and Procedures, Part D Coverage of Anti-Obesity Medications (AOMs) and Application to the Medicaid Program, and Ensuring Equitable Access to Medicare Advantage Services— Guardrails for Artificial Intelligence (AI). CMS, however, does want to acknowledge the broad interest in regulation of AI and will continue to consider the extent to which it may be appropriate to engage in future rulemaking in this area.

This is a big (correct) step. For all the hype surrounding the use of AI in health care, there remain many questions, key among them: How is it actually used? Should it be regulated, and if so, how? While American Action Forum President Doug Holtz-Eakin aptly pointed out that a Friday post-market closure rule finalization seems to be an administration “best practice,” CMS did seem to get something right in soft-pedaling the regulation of AI. And though it did prompt some surprise among the industry, it was the right call.

According to the consulting firm Accenture, the application of AI in clinical scenarios – including robot-assisted surgery, virtual nursing assistants, and administrative workflow assistance – has the potential to generate $150 billion in annual savings for the U.S. health care economy by 2026. But some organizations still can’t ascertain the actual return on investment (ROI) for implementing AI. Only 33 percent of payers, 17 percent of health systems, and 18 percent of health services/tech companies have a documented positive ROI. While that shouldn’t have any bearing on those of us not employed at those companies, it does indicate that – since the ROI for this technology is unclear (as is its adoption) – we should probably be careful about creating a regulatory structure for AI implementation since we can’t fully comprehend or anticipate our current and future use-cases.

In general, people show positive attitudes when asked about the use of AI in health care. Approximately 59 percent of Americans believe that using AI will lead to better health outcomes through assisting diagnosis and treatment. In the same survey, a strong majority of people (77 percent) felt that AI can work to effectively reduce health care disparities. And there are areas where AI-assisted health care is positively impacting patient care and accessibility. Take radiology, for example. AI-powered diagnostic tools can analyze imaging scans faster and with similar or better accuracy than radiologists, leading to faster diagnoses. The Mayo Clinic, in collaboration with Google Health, found that AI applications in radiation therapy planning improved efficiency and precision, potentially accelerating treatment initiation.

But according to the American Medical Association, physicians are still limiting how they use AI in practice. For example: 12 percent of physicians used AI to assist in diagnosing patients; 20 percent used AI to aid in creating discharge instructions, care plans, and progress notes; and 10 percent use it for patient-facing health recommendations and care engagement. While these proportions will grow as better, smarter AI models are developed, we have a long runway to decide if there is any regulation necessary for these use-cases.

The potential benefits of AI in health care are compelling, but any regulatory framework must be carefully considered to prevent premature restrictions – along with ensuring patient safety isn’t at risk. By not jumping reflexively to finalize the proposed rule, CMS will be better able to navigate the provider landscape and thoughtfully ensure that rulemaking doesn’t create implementation issues.

Chart Review: Addressing the Biosimilars Pipeline Void Could Save Billions

Nicolas Montenegro, Health Policy Intern

The biologics/biosimilars industry – which produces large molecule pharmaceutical drugs that are typically used to treat complex chronic conditions such as cancers and autoimmune disorders – is expected to grow significantly over the next decade, according to a recent IQVIA report. Since the Food and Drug Administration first issued guidance on biosimilars in 2014, 73 biosimilars have been approved. The projected growth can be attributed to two main factors: (1) the need to address rising prevalence of chronic conditions, and (2) the loss of exclusivity (LoE) for many high-profile biologic patents. Over the next 10 years, 118 biologic patents are anticipated to expire, presenting a $232 billion opportunity for potential biosimilar competitors. The chart below demonstrates the market share of these biologics facing patent LoE as a portion of the total projected biologic market size from 2025 to 2034. Biosimilars are forecasted to expand at an annual growth rate of roughly 17 percent, reaching total revenues of over $93 billion by 2034.

The Department of Health and Human Services has studied the success of biosimilars in improving access to critical medicines and reducing drug spending in the health care system. The introduction of biosimilars has saved at least $36 billion from 2015 (when the first biosimilar entered the market) to 2023, with $4.4 billion saved for Medicare Part B in 2023 alone. While these IQVIA estimates point to the possibility of substantial savings in the health care sector – as biosimilars provide increased access and affordability to patients through savings on the wholesale cost of the reference product, with savings as high as 87 percent – achieving these benefits is not so simple. Of the 118 biologics losing patent exclusivity in the next decade, about 90 percent do not currently have biosimilar alternatives in development. The shrinking biosimilar pipeline is likely due to myriad external and internal factors, including high investment costs, skewed concentrations of biologics by therapeutic class, regulatory frameworks, reimbursement discrepancies, and uncertainty related to the Inflation Reduction Act. Policymakers may look to bolster industry pipelines to ensure continued access to these important drugs.