UnitedHealth nH Predict Denied Medicare Post-Acute Care Without Clinician Authorization
AI coverage-denial system allegedly committed post-acute care denials to Medicare Advantage beneficiaries without named clinician authorization; the class action alleges — as a lawsuit claim, not an established finding — a 90% reversal rate on appeal.
What happened
A class action filed in November 2023 alleges that UnitedHealth and NaviHealth used the nH Predict AI model to override treating physicians and deny post-acute care to Medicare Advantage members, and that roughly 90% of the model's denials were reversed when members appealed — a figure that remains a lawsuit allegation, not an established finding. The complaint further charges that no named clinician signed each denial and no audit record existed of who approved the decision or on what basis. The federal court allowed breach-of-contract claims to proceed and ordered broad discovery (2025-2026); only approximately 0.2% of affected members appealed, allowing erroneous denials to stand as final.
What the agent did
According to the class action complaint, the nH Predict system and coverage-decision workflow committed denials of post-acute care benefits to Medicare Advantage members without requiring prior authorization from a treating clinician or any named human approver, with no audit trail of who approved each decision.
The irreversible effect
Thousands of Medicare Advantage beneficiaries were denied post-acute care benefits, with denials becoming final unless the member appealed; the lawsuit alleges denial of medically necessary care, economic and health harm to vulnerable beneficiaries, and erosion of trust in automated coverage decisions.
Root cause
No segregation of duties between the system recommending denials and the system finalizing them; no requirement for a named human clinician to approve each denial before it became irreversible; absence of audit accountability (no record of who authorized each decision); proxy access allowing denials to bypass approval gates.
How a maker-checker control would have refused it
MakerChecker would block this by denying the assess-and-deny workflow access via two complementary refusals: (1) skill_not_granted—the assessment system holds no commit skill and cannot finalize denials; (2) high_risk_requires_gate—even a clinician reviewer cannot commit through the proxy and must use a governed approval-gate flow where a named clinician signs each denial before execution. The audit trail records who approved what and on what basis, leaving no room for unsigned denials to execute.
Runnable reproduction
This incident ships as a runnable scenario in the open-source repository. Point the enforcement engine at the policy and watch the action get refused, with the refusal written to a signed audit record.
examples/unitedhealth-nhpredict-ai-medicare-denials
Primary sources
- CBS News: UnitedHealth Lawsuit AI Deny Claims Medicare Advantage Health Insurance Denials
- Healthcare Finance News: Class Action Lawsuit Against UnitedHealth's AI Claim Denials Advances
- ArentFox Schiff Alert: Federal Court Orders Broad Discovery Against UHC AI Coverage Denial Lawsuit
- MakerChecker Insights: UnitedHealth nH Predict AI Medicare Denials
Accuracy and corrections
This entry describes a publicly reported incident and is compiled from the primary sources listed above. Where an account is a legal allegation rather than an established finding, the entry labels it as such. Summaries can still contain errors. If you can document a correction, email hello@makerchecker.ai and we will review and correct it, with the change noted, within 14 days.
See it for yourself
Reading is one thing. Watch it block an agent.
One command starts the demo: an agent stopped from signing off its own work, and the signed evidence file an inspector can check for themselves.
Designed against the rules your auditors already enforce.