Share on FacebookShare on TwitterShare on LinkedIn
In the realm of healthcare, there lacks a centralized authority overseeing the regulation of algorithms. While certain commercial tools like the heart rate monitor embedded in the Apple Watch receive approval from the Food and Drug Administration (FDA), algorithms crafted by electronic health records (EHR) behemoths such as Epic or Oracle Cerner fall within the purview of the Office of the National Coordinator for Health IT (ONC).
Micky Tripathi, the director of ONC, elucidated to Forbes the essence of the agency’s freshly minted regulations slated to come into effect on March 11, 2024. Rather than directly regulating AI-powered tools, the approach mirrors a “nutrition label” akin to those found on food products. EHR companies are mandated to delineate 33 distinct data elements, furnishing hospitals and physicians with insights into the model’s training data and functionality. Referred to as model cards in technical jargon, these labels are geared towards empowering customers to make informed decisions, steering clear of subjective judgments on product quality. Tripathi underscored the broad applicability of these regulations, transcending the scope of FDA oversight.
While the regulatory ambit currently encompasses tools developed by EHR entities, Tripathi aspires for app developers and health systems fashioning in-house solutions to also embrace the concept of nutrition labels. Emphasizing the significance of transparency, he noted that vacant fields in the labels could be as enlightening as filled ones. Customers are encouraged to exercise discretion if developers opt to omit information pertaining to crucial aspects like health equity, safety, or performance bias. The overarching objective, as Tripathi articulated, is not to proscribe methodologies but to instigate a structured approach towards algorithm integration within products.
