Get Full Government Meeting Transcripts, Videos, & Alerts Forever!
Get email alerts on the Ai Prior Authorization topic
No spam. Unsubscribe anytime.
Lawmakers and experts warn against AI‑driven prior‑authorization programs, cite CMS 'WISER' pilot
Summary
Witnesses and members raised concerns that using AI to automate prior‑authorization reviews could delay or deny medically necessary care, questioned incentive structures that pay vendors from savings tied to denials, and urged human‑in‑the‑loop standards and independent validation.
Members of the House Energy and Commerce hearing focused extensive questioning on the use of AI in prior authorization and on a CMS pilot program discussed in testimony.
Michelle Mello, professor of law and health policy at Stanford, told the committee that prior authorization is already “fraught” and that adding AI could make outcomes better or worse, but that the effect is unknown without independent study. “We don't know because there are no publicly available information that would enable somebody like me to be able to tell you, does using AI make prior authorization better for patients or worse?” she said.
Multiple members and witnesses described the administration’s pilot program — referred to in testimony as a WISER pilot — that would extend use of automated reviews to parts of Medicare. Lawmakers and witnesses flagged two core concerns: (1) financial incentives tied to denying claims and (2) the lack of public performance data and standards for the human review that follows an algorithmic recommendation.
Andrew Toye, CEO of Clover Health, said his company does not use AI to make coverage denials and that “AI should not be used to review, to deny in any case.” Toye and others told members that if AI is used to speed administrative work, human clinicians must remain the final decisionmakers and peer‑to‑peer clinician reviews should be available when denials occur.
Witnesses recommended specific policy responses: require disclosure of when AI is used in coverage decisions; mandate independent, post‑deployment evaluation of algorithms; define what “human review” must include (for example, a fresh, unprimed clinician review rather than a cursory sign‑off on a packaged denial); and extend reporting and oversight to commercial employer‑sponsored plans that are currently less transparent.
Members pressed CMS and the committee for immediate oversight and asked for clearer statutory authority and modernized review processes at FDA and CMS so agencies can evaluate software products more effectively before they influence coverage and care.
