AI is deciding whether or not people receive medical care

Major health insurers and the federal Medicare program are using AI systems to vet and often deny medical claims and prior authorizations, and those practices are now the subject of multiple lawsuits and clinician complaints. Plaintiffs allege UnitedHealth’s nH Predict and similar tools at Humana and Cigna generate specific predictions about patients’ needs that are opaque to patients and doctors, drive rapid cutoffs of nursing-home and rehab care, and sometimes replace meaningful human review. Medicare has rolled out its own AI tool for prior authorizations, and clinicians report patients left without needed medicines or in tears when coverage is denied. Legal filings argue that employees are discouraged from overruling algorithmic recommendations and that insurers hide the logic behind proprietary models, while insurers point to regulations requiring human sign-off.

The complaints include detailed examples: Gene Lokken, an elderly rehab patient whose nursing-home coverage was allegedly stopped after about two weeks and who later died, plus other patients forced to pay thousands or left with unresolved medical needs. Cigna is accused of batch-denying claims based on preset criteria; Humana faces similar charges tied to nH Predict. Experts warn these models miss nonnumeric, context-sensitive factors and can erode clinicians’ judgment. Practical advice includes appealing denials, protecting vulnerable people, and pressing lawmakers to regulate AI use in health care.