Medicare’s AI-Powered WISeR Program Faces Criticism Over Delays and Testing

Newly released government records indicate that a Medicare initiative using artificial intelligence (AI) for reviewing prior authorization requests has resulted in extended treatment delays and raised concerns about inadequate testing. The records came to light through a Freedom of Information Act (FOIA) lawsuit filed by the Electronic Frontier Foundation (EFF). This provides a detailed overview of the Wasteful and Inappropriate Service Reduction (WISeR) model, a Medicare program launched in January relying on AI-assisted reviews for specific medical procedures.

The WISeR Initiative

The WISeR program represents one of the federal government’s most significant efforts to integrate AI into Medicare administration. It is currently in operation across six states: Texas, Arizona, New Jersey, Ohio, Oklahoma, and Washington. The program targets 13 medical services identified by the Centers for Medicare & Medicaid Services (CMS) as particularly susceptible to fraud, waste, or misuse. Providers must secure authorization before treatment, utilizing technology platforms managed by private contractors that employ AI tools to assess requests.

Documents released through the EFF lawsuit highlight widespread operational issues since the program’s inception. Key concerns include “widespread delays” in the responses to authorization requests, which contradict CMS’s expectation of vendors responding within 72 hours. EFF noted that requests have gone unanswered for up to 83 days, leading to inappropriate denials and reports of patient harm. Kevin Thompson, CEO of 9i Capital Group, commented, “The systems were not thoroughly tested prior to launch, causing lengthy delays beyond the 72-hour prior authorization turnaround.”

Testing Concerns and Delays

“AI-driven prior authorization determinations can lead to unwarranted—and even discriminatory—delays or denials of necessary medical care,” EFF stated last week. They emphasized that rigorous testing is essential to prevent such issues.

Questions remain regarding the thoroughness of the AI systems’ testing before deployment. EFF’s FOIA request included records related to testing for accuracy, bias, and other abnormalities, along with audits and evaluations of vendors involved. Although CMS maintains that a qualified human clinician reviews all denials, research indicates AI recommendations often influence human decisions.

Michael Ryan reflected on the situation, noting that inefficient processes persist when authorization requests intended for swift responses drag on for extended periods. “For the beneficiary, it is delayed treatment,” Ryan stated.

Future Implications

CMS has acknowledged the concerns and promised improvements. “CMS is firmly committed to ensuring timeliness, accuracy, and transparency under the WISeR model,” it declared on its WISeR FAQ page, pledging partnership with providers and participants to swiftly identify and resolve issues.

Previous findings echo similar complaints voiced by doctors and patients in the program. A June investigation by KFF Health News found long waits and confusion following WISeR’s integration into traditional Medicare. Providers reported needing to adapt quickly to new administrative demands.

Despite these hurdles, Ryan does not anticipate Medicare shelving AI technologies. “WISeR is scheduled through 2031, and CMS clearly sees technology-assisted prior authorization as part of the future,” he said. Expanding the program without demonstrating its ability to consistently improve care would be premature.