AI Will Soon Have a Say in Approving or Denying Medicare Treatments (kffhealthnews.org)

The Biden-era administration announced a new Medicare pilot called WISeR (Wasteful and Inappropriate Service Reduction) that will start Jan. 1 and run through 2031 in six states (Arizona, Ohio, Oklahoma, New Jersey, Texas and Washington). The program will test an AI algorithm to screen certain outpatient services—initially skin and tissue substitutes, electrical nerve stimulator implants, and knee arthroscopy—for potential “low-value” or fraud-prone care and trigger prior authorization reviews. CMS says automated determinations will be subject to human clinician review and vendors cannot be paid based on denial rates, though vendors may receive shared-savings rewards for reducing costs. For the AI/ML community, WISeR is significant because it represents Medicare’s first large-scale experiment delegating utilization decisions to algorithmic tools, raising questions about model design, evaluation, transparency, and perverse incentives. Technical and policy concerns include ambiguous definitions of “meaningful human review,” subjective outcome measures, reliance on contractor self-assessment, and the risk that cost-focused models systematically deny high-cost or complex care. Prior scrutiny (ProPublica, AMA surveys, class-action suits) shows rushed or opaque reviews can harm patients; proponents argue AI could speed up a cumbersome process and reduce waste. The pilot will be a critical test of how to govern, audit, and validate ML systems in high-stakes reimbursement workflows.