344
Productivity & Workflow355
Automation & Workflow224
Software Development251
Marketing & Growth192
AI Infrastructure & MLOps174
Writing & Content Creation203
Data & Analytics141
Design & Creative170
Photography & Imaging156
Customer Support131
Sales & Outreach125
Voice & Speech135
Education & Learning131
Operations & Admin87
A new CMS test uses AI to flag certain Medicare services for extra review. Supporters say it could cut waste, and critics warn of delays and denials.
In short: The US government is testing an AI-assisted system to review some Medicare coverage decisions in six states.
The Centers for Medicare and Medicaid Services, or CMS, has started a demonstration project called WISeR, short for Wasteful and Inappropriate Service Reduction Model. It runs through December 2031 in six states. The goal is to reduce unnecessary medical spending in original Medicare.
WISeR uses AI, including machine learning (software that looks for patterns, like a spam filter for medical bills), plus human clinical review. CMS says it will focus on services it believes are vulnerable to overuse, fraud, or abuse. Examples mentioned include some skin and tissue substitutes, electrical nerve stimulator implants, and certain knee procedures.
Prior authorization is the pre-approval step many patients know from private insurance. It can help control costs, but it can also delay care. Critics worry that adding AI could lead to more wrong denials, meaning care a doctor thinks is needed gets rejected or slowed.
A 2025 American Medical Association survey found that 61 percent of doctors who responded were concerned AI would make denials worse. Federal reports have also raised concerns about prior authorization in Medicare Advantage, where many denials are later overturned on appeal.
If WISeR expands or shapes future policy, more people in original Medicare could face a process that feels like getting permission before treatment. That could reduce waste, but it could also add delays, paperwork, and appeals when time matters.
Source: Arstechnica