In a race against time, U.S. health systems are working to ensure their clinical algorithms comply with new federal anti-discrimination regulations. The upcoming May deadline requires federally funded health institutions to vet their decision-making tools for potential bias based on protected traits such as race and sex.
The challenge lies in the widespread use of calculators and algorithms that incorporate these characteristics to predict patient outcomes and guide treatment decisions. While some race-based tools have been phased out over the past four years, many others remain in a gray area, leaving health systems grappling with how to balance compliance and clinical best practices.
Rohan Khazanchi, an internal medicine and pediatrics resident physician in Boston, highlights the dilemma: “There are many algorithms that still sit in this gray area. We don’t really have an alternative, and we’re sitting here going, OK, how do we both comply with the rule and still adhere to what we think is clinical best practice?”
The new regulations, part of the Affordable Care Act’s Section 1557, aim to prevent discriminatory practices in healthcare. They require health systems to conduct “reasonable diligence” in identifying potentially biased outputs from clinical algorithms and AI tools.
This push for equity in healthcare decision-making tools reflects a growing awareness of how embedded biases can perpetuate health disparities. As the deadline approaches, health systems must navigate the complex task of ensuring their algorithms promote fair and equitable care while maintaining clinical efficacy.
The outcome of this nationwide effort could significantly impact how medical decisions are made and potentially reshape the landscape of healthcare delivery for minority communities across the United States.
See: “Health systems race to show clinical algorithms don’t discriminate” (January 14, 2025)
