AI Is Already Influencing Decisions About Medical Care

Summary

Artificial intelligence is already affecting whether some people receive or continue receiving medical care, according to this Vox report. UnitedHealth, Humana, and Cigna face lawsuits alleging that AI systems contributed to wrongful denials of nursing-home, hospital, rehabilitation, and diagnostic care. The UnitedHealth and Humana complaints focus on nH Predict, which plaintiffs say employees followed too closely; UnitedHealth disputes the allegations and says federal rules require a human doctor to make the final decision. The Cigna lawsuit alleges that another system enabled doctors to deny claims in batches, potentially bypassing individualized review. The report also examines Medicare’s WISeR pilot, launched in six states to use AI and machine learning alongside clinical review for prior authorization covering about a dozen procedures. Doctors and patients reported denials, processing delays, and at least one payment glitch; documents obtained by the Electronic Frontier Foundation included a case that took 83 days despite a stated 72-hour target. One company using the model reportedly denied more requests than it approved, while physicians described patients calling in pain because treatment was delayed. The article says traditional algorithms have long been used in insurance, but newer systems make more specific predictions about future health while obscuring the data, reasoning, and degree of human involvement behind decisions. Researchers and advocates warn that social and clinical circumstances may not be captured by numerical models. States and Congress are considering transparency and human-review requirements, but regulators have not established a verifiable definition of meaningful human oversight. Until broader rules develop, experts advise patients to appeal denials, support vulnerable people, and press officials for stronger oversight. The article presents the claims and allegations as contested litigation and reporting, not as settled findings.