The memo from the Centers for Medicare & Medicaid Services (CMS) to Medicare Advantage insurers emphasizes that algorithms or artificial intelligence should not be used to determine care or deny coverage to members on Medicare Benefit plans. This directive follows recent lawsuits against UnitedHealth and Humana regarding the use of an AI-powered tool called nH Predict, which allegedly led to improper denial of care to elderly patients on MA plans.
The nH Predict tool, created by NaviHealth and utilized by both carriers, is under scrutiny for providing projections on the duration of post-acute care needed by patients, often conflicting with physicians’ recommendations and Medicare coverage guidelines. Employees using nH Predict are reportedly disciplined for deviating from its estimates, leading to delayed payments for services such as skilled nursing care.
Despite its large data pool of 6 million individuals, nH Predict is criticized for not considering comprehensive patient-specific factors. The CMS highlights the importance of individual patient conditions in decision-making, stating that algorithms should not replace medical records, physician input, or specific patient circumstances.
Insurers are permitted to use AI tools for compliance checks but must ensure that coverage decisions align with regulations. The CMS warns against altering coverage criteria using AI and stresses the need for transparency in denial explanations. Concerns about bias and discrimination in AI systems are also addressed, urging insurers to prevent the perpetuation or introduction of biases.
In response to potential non-compliance, the CMS plans to increase monitoring of MA plans to ensure adherence to Medicare laws and internal protection criteria. Violations may result in penalties, corrective actions, and sanctions on membership and marketing activities.
