A Federal Judge Demands Transparency on Claims-Denying AI
A Minnesota federal judge dealt a major blow to UnitedHealth Group on March 9, 2026. The court ordered the health insurance giant to hand over a broad range of internal documents related to the AI algorithm it uses to manage Medicare Advantage claims. Specifically, the judge ruled that UnitedHealth must disclose records detailing whether its technology was designed to override the clinical judgment of doctors. This ruling marks a pivotal moment in the ongoing legal battle over algorithmic accountability in healthcare.
Key Details of the Order
The court largely sided with plaintiffs in their motion to compel discovery. Judges granted or partially granted requests across six of seven document categories. This decision gives plaintiffs’ attorneys direct access to internal records that UnitedHealth had previously shielded from scrutiny.
What Is the nH Predict Algorithm?
The Tool at the Center of the Controversy
The lawsuit centers on nH Predict, an AI tool developed by naviHealth, a care management company that Optum — a UnitedHealth subsidiary — acquired in 2020. The company later rebranded naviHealth as Home & Community Care in 2024. Plaintiffs allege that nH Predict routinely overrode physicians’ decisions, leading to premature denials of medically necessary post-acute care for elderly Medicare Advantage members.
The 90% Error Rate Allegation
One of the most striking allegations in this case involves the tool’s accuracy. According to plaintiffs, nH Predict carries a 90% error rate — meaning that nine out of every ten denied claims were ultimately reversed on appeal. Despite this, plaintiffs argue, UnitedHealth continued using the tool because only about 0.2% of policyholders ever appeal a denied claim. For the insurer, the math reportedly worked in its favor financially, even as patients suffered.
UnitedHealth has disputed these characterizations. An Optum spokesperson previously told media outlets that nH Predict does not make coverage determinations. Instead, Optum maintains, it serves as a guide to help inform providers, families, and caregivers about a patient’s likely care needs. Coverage decisions, the company insists, rest on CMS criteria and each member’s individual plan terms.
What Documents UnitedHealth Must Produce
A Wide-Ranging Discovery Order
Under the court’s March 9 order, UnitedHealth must produce an extensive set of documents. The timeline covers records dating back to January 2017, well before nH Predict’s 2019 deployment. The court rejected UnitedHealth’s argument that pre-2019 records were irrelevant.
The required documents include:
- Internal policies and procedures for post-acute care claims adjudication
- All records analyzing or discussing nH Predict’s design and function
- Documents related to UnitedHealth’s acquisition of naviHealth, specifically those connecting the deal to post-acute care cost savings
- Records concerning any government investigations into the company’s AI use in claims decisions
- Performance evaluations and compensation records for post-acute care coordinators and medical directors
- Documents related to the company’s internal AI review board, including board member identities
- Names and contact information for medical directors and care coordinators involved in denials for 300 members of the proposed nationwide class action
Why the Scope Matters
The breadth of this order is significant. Consequently, plaintiffs will gain insight not only into how nH Predict functions today but also into the business motivations that may have shaped its development. Furthermore, records linking the naviHealth acquisition to cost-saving strategies could prove especially damaging if they suggest the tool’s primary purpose was financial rather than clinical.
How the Case Reached This Stage
Origins of the Lawsuit
The original complaint dates to November 2023. Families of two deceased Medicare Advantage members filed it in the U.S. District Court for the District of Minnesota. The plaintiffs, representing a proposed nationwide class, alleged that UnitedHealth illegally denied elderly patients medically necessary post-acute care. They claimed the AI tool was the real decision-maker — not physicians.
February 2025: A Partial Green Light
In February 2025, Judge John R. Tunheim dismissed several counts, including claims for unjust enrichment and bad faith insurance practices, finding them preempted by the federal Medicare Act. However, the judge allowed breach of contract and breach of the implied covenant of good faith and fair dealing claims to proceed. Notably, he also waived the requirement that plaintiffs exhaust all administrative remedies first, calling UnitedHealth’s appeal process effectively “futile.”
September 2025: Discovery Dispute Begins
Subsequently, UnitedHealth sought to narrow and delay discovery. The court denied that request in September 2025, finding that splitting discovery phases would cause unnecessary delays. The stage was therefore set for the March 2026 compel order.
Broader Industry Implications
Other Insurers Face Similar Scrutiny
UnitedHealth is not alone in facing legal pressure over AI-driven claim denials. Cigna faced a lawsuit alleging its PXDX algorithm allowed doctors to deny claims in large batches without individual review. Humana faced similar accusations in court. Together, these cases signal a broader reckoning for the health insurance industry’s use of automated decision-making tools.
What This Means for Medicare Advantage Members
For the millions of Americans enrolled in Medicare Advantage plans, this ruling carries real weight. It suggests that insurers cannot simply invoke trade secrecy to shield their algorithms from legal scrutiny. Moreover, it reinforces a growing legal consensus: when an insurer’s own plan documents promise physician-led review, replacing that review with an AI tool may constitute a breach of contract.
What Comes Next
Discovery and the Road to Trial
With discovery now compelled, plaintiffs’ attorneys will begin reviewing the documents UnitedHealth produces. This process could reveal internal communications, financial projections, and design specifications that directly address whether nH Predict was built to limit care for economic reasons. Additionally, the identities of medical directors and care coordinators now subject to disclosure could expand the circle of accountability.
A Test Case for AI Accountability in Healthcare
Ultimately, this lawsuit has become one of the most closely watched test cases for AI accountability in U.S. healthcare. Regulators, advocacy groups, and competing insurers are all watching to see how the evidence unfolds. The outcome could reshape how insurers deploy and document AI tools — and what transparency courts will demand from them going forward.
