Federal government proposes AI ‘guardrails’ for Medicare Advantage plans

The Centers for Medicare and Medicaid Services have proposed “guardrails” to ensure that the use of artificial intelligence for Medicare Advantage insurance plans does not result in inequitable access to health care-related services.

See also: Maximize the utility of data in mission execution, citizen services and education

The AI ​​offering, which is part of a broader whole proposed rule for “technical changes” to Medicare Advantage programs for 2026, specifies that AI or automated systems “must be used in a manner that preserves equitable access” to Medicare Advantage services and that insurers must “remove unnecessary barriers to care resulting from the use of inappropriate prior data”. authorization by clarifying requirements for the plan’s use of internal coverage criteria.

The proposal supports an October 2023 Biden administration executive order that directs agencies to ensure that AI tools do not hinder the advancement of equity and civil rights and that the use of AI within healthcare organizations “does not negate equal opportunity and Justice for the American People,” CMS said in an information sheet (see: Biden’s Executive Order on AI: What Are the Benefits for Healthcare?).

“Given the growing use of AI in healthcare, we believe it is necessary to ensure that the use of AI does not result in unfair treatment or bias – or both – within the health care system, but rather be used to promote equitable access to care and culturally competent care for all enrollees,” the provision states.

Some Medicare Advantage plans’ use of AI is already facing scrutiny. For example, the estates of two deceased UnitedHealthcare Medicare Advantage policyholders filed a proposed federal class-action lawsuit last year alleging that the insurance company uses an AI tool to illegally deny necessary coverage for post-acute care , including skilled nursing care and home care, for elderly participants (see: Lawsuit: Health insurer’s AI tool “illegally” denies claims).

Rather than imposing entirely new requirements, the CMS proposal emphasizes mandates that are already part of existing Medicare Advantage regulations.

“We are not proposing any regulatory changes to these requirements because these existing requirements already apply to MA plans if they use AI or automated systems,” the proposal states.

Instead, CMS said it reiterates that in the event a Medicare Advantage plan uses AI or automated systems, it must comply with regulations that prohibit a Medicare Advantage organization “from denying, limiting, or conditioning coverage or provision of benefits to persons eligible to enroll.” in an MA plan proposed by the organization based on any factors related to the health condition.

As the healthcare system evolves and uses new and emerging AI tools, CMS said it “feels the need to clarify that these tools, including but not limited to, machine learning, patient care decision support tools, and/or other algorithmic tools, must not violate CMS rules.”

If Medicare Advantage organizations use these AI tools or automated systems in any way, “it is their responsibility to ensure that the use of these tools complies with all existing Medicare policies, including, but not limited to, providing culturally competent care to all enrollees in a non-discriminatory manner. »

The proposed rule is expected to be published in the Federal Register on December 10.

But once President Donald Trump takes office on January 20, 2025, it will be up to the new administration to decide whether to revoke Biden’s executive orders and other pending proposed regulations and rules.