On December 4, 2024, a gunman shot UnitedHealthcare CEO Brian Thompson on a Manhattan sidewalk outside a shareholder conference. The public reaction was unlike anything in modern American history: millions of Americans expressed anger not at the perpetrator but at the insurance system Thompson represented. Social media filled with stories of denied claims, overturned treatments, and the exhausting process of fighting a faceless bureaucracy for care that a doctor had already ordered. The fury shocked the political class but did not surprise anyone who works in patient advocacy, emergency medicine, or insurance regulation. It had been building for years — one denial letter at a time — as Americans discovered that the entity deciding whether they could receive surgery, rehabilitation, or specialist care was increasingly not a physician. It was an algorithm. Source: The Guardian, December 5, 2024; NBC News, December 2024.

This is not a conspiracy theory. It is documented fact — investigated by ProPublica, reported by STAT News, examined by the U.S. Senate Finance Committee, and now being challenged by state regulators, federal rulemakers, and class action plaintiffs across the country. Health insurers — including the largest in America — have built AI-powered systems that generate denial recommendations in seconds, review claims in batches without reading individual patient files, and systematically flag treatments for rejection based on statistical population models rather than clinical evaluation of the person sitting in a hospital bed. Here is what the evidence actually shows, which systems are documented, what the regulatory landscape looks like in April 2026, and — most critically — the precise six-step process that gives you the best chance of reversing a wrongful denial.

How Insurance AI Denial Systems Actually Work — And Why They Are Built This Way

Insurance companies use a range of terms for these tools: utilization management technology, algorithmic prior authorization, clinical decision support systems. The architecture is similar across implementations. An AI model — trained on millions of historical claims, diagnostic codes, treatment protocols, and payer coverage policies — ingests a patient's diagnosis code (ICD-10 code), age, geographic location, insurance plan tier, and treatment history and generates a recommendation in seconds. The recommendation is typically binary: approve or deny. The question is whether a human being with clinical judgment reviews that recommendation before it determines a patient's care — and the documented evidence suggests that at scale, that review is often nominal at best. Source: ProPublica, March 2023; STAT News, November 2023.

The business logic is not difficult to understand. A major U.S. health insurer processes millions of prior authorization requests per year. UnitedHealthcare alone manages coverage for approximately 50 million Americans. Cigna covers approximately 18 million. Aetna/CVS covers approximately 24 million. At those volumes, meaningful clinical review of every prior authorization request would require thousands of physicians working full-time exclusively on authorization decisions — a cost that would directly reduce profitability. Algorithmic systems reduce that cost dramatically by pre-screening claims and generating denial recommendations that humans then rubber-stamp or reverse in bulk. The problem — documented in the investigations below — is what happens when the rubber-stamping becomes the primary function and the clinical review becomes a checkbox. Source: Kaiser Family Foundation Health Insurance Coverage data, 2024; ProPublica, March 2023.

The Cigna PXDX Investigation: 1.2 Seconds Per Claim

The most thoroughly documented case of AI-assisted mass claim denial is Cigna's PXDX system, exposed by a ProPublica and The Capitol Forum investigation published in March 2023. PXDX stands for Procedure to Diagnosis — a matching system that compares a requested procedure code against a patient's diagnosis code and automatically flags the request for denial if the pairing does not match Cigna's internal coverage criteria. The investigation obtained internal Cigna documents and conducted interviews with former employees who described the practical operation of the system in detail. Source: ProPublica and The Capitol Forum, "How Cigna Saves Millions by Having Its Doctors Reject Claims Without Reading Them," March 25, 2023.

According to the ProPublica investigation, Cigna physicians were presented with batches of PXDX denial recommendations through an internal review interface and approved those denials in bulk — clicking through dozens of cases per session without opening individual patient files, reviewing medical records, or examining the treating physician's clinical rationale. The investigation found that over a two-month period reviewed from internal records, Cigna physicians reviewed and approved the denial of approximately 300,000 patient requests, with an average review time of approximately 1.2 seconds per case. The investigation concluded that this time was insufficient to open and read an individual patient file — meaning the denials were being approved based solely on the algorithmic flag, without independent clinical review of each patient's specific situation. Source: ProPublica and The Capitol Forum, March 25, 2023.

Cigna disputed the characterization, arguing that the PXDX system was appropriately used for a specific, limited category of routine denials where treatment was clearly outside coverage criteria and did not involve medical judgment — cases where the denial was essentially administrative. The company argued that for complex clinical cases, a separate process applied. Critics, including patient advocates and the physicians interviewed by ProPublica, countered that no process was reliably ensuring that algorithmically flagged cases were being sorted into the appropriate category, and that the batch review system created structural pressure toward denial regardless of individual circumstance. The California Department of Insurance subsequently opened an inquiry. No federal enforcement action had been announced as of the investigation's publication. Source: ProPublica, March 25, 2023; Cigna response, March 2023.

UnitedHealthcare's nH Predict: When an Algorithm Overrides Your Doctor

The second thoroughly documented case involves Medicare Advantage — the private insurance version of Medicare that covers approximately 32 million Americans over 65, a number that has roughly doubled over the past decade. STAT News reported in November 2023 that UnitedHealthcare, through its subsidiary naviHealth, used an AI tool called nH Predict to generate coverage determinations for post-acute care — the rehabilitation and skilled nursing care that patients typically need after major hospitalizations for procedures like hip replacements, strokes, and cardiac events. Source: STAT News, "UnitedHealth Used AI to Deny Coverage, Lawsuit Says," November 14, 2023.

The nH Predict system generated a predicted length of stay based on population-level statistical modeling — essentially, how long an average patient with a given diagnosis profile typically remains in post-acute care. naviHealth case managers used that prediction as the basis for determining when UnitedHealthcare coverage would end for a Medicare Advantage patient's rehabilitation stay. Critically, according to STAT News and the class action lawsuits filed subsequently, the algorithm's predicted length of stay was being used to cut off coverage even when the patient's treating physician had determined that continued care was medically necessary. The patient's physician recommendation — the clinical judgment of the professional directly evaluating the patient — was being systematically overridden by an algorithm that had never examined the patient and was not making a clinical determination about that patient's individual condition. Source: STAT News, November 2023; class action complaints filed in U.S. District Court, 2024.

Federal class action lawsuits filed in 2023 and 2024 alleged that UnitedHealthcare and naviHealth violated the terms of Medicare Advantage coverage by using the nH Predict algorithm to deny care that traditional Medicare — which Medicare Advantage is legally required to cover at equivalent levels — would have approved. The lawsuits alleged that the algorithm systematically generated premature coverage cutoffs and that a significant portion of coverage decisions challenged through the appeals process were ultimately overturned, suggesting the initial algorithmic determinations were frequently erroneous. UnitedHealthcare and naviHealth disputed the characterizations and denied that nH Predict was used to replace clinical judgment. As of 2026, litigation in this area is ongoing. Source: STAT News, November 2023; federal court filings, 2023–2024.

What the Senate Found: A Systematic Pattern Across the Industry

The ProPublica and STAT News investigations documented specific systems at specific companies. The U.S. Senate Finance Committee's October 2022 investigation — which predated both media investigations — documented a systematic industry-wide pattern. The committee examined prior authorization data from the three largest Medicare Advantage insurers: UnitedHealthcare, Humana, and CVS Health/Aetna. Combined, these three companies cover the majority of Medicare Advantage enrollees in the United States. Source: U.S. Senate Finance Committee, "Identifying Vulnerabilities and Ensuring Integrity in Medicare Advantage," October 2022.

The committee's central finding was that these plans denied prior authorization requests at substantially higher rates than traditional Medicare for comparable procedures and services — in some categories, at rates roughly five times as high. The committee's report identified the use of algorithmic and automated prior authorization tools as a factor contributing to elevated denial rates and raised concerns about whether the algorithms were properly calibrated to reflect Medicare coverage standards, as legally required. The committee specifically noted that the high rates of successful appeals in Medicare Advantage — meaning patients who were denied and fought back successfully — were evidence that the initial denials were frequently incorrect on the clinical merits. A denial that gets overturned on appeal was almost certainly wrong in the first place. Source: Senate Finance Committee, October 2022.

The Scale of the Problem: What the Numbers Show Across All Plans

The number that deserves the most attention in that table is the final row. Fewer than 1 in 1,000 Americans who receive a denied claim ever files a formal appeal — and of those who do, more than 40% succeed. These two numbers together define an enormous injustice of inaction. For every American who successfully appealed a wrongful denial and received their coverage, there are roughly 1,500 people who received the same wrongful denial, never appealed, and either went without care, paid out of pocket, or delayed treatment with potential health consequences. The appeal process works. The overwhelming majority of Americans never use it. Source: KFF Health News analysis, 2023.

Why don't more people appeal? The American Medical Association's 2022 prior authorization survey — which surveyed physicians, not patients — found that 94% of physicians reported that prior authorization requirements had delayed care for their patients, and 34% reported that prior authorization delays had led to a serious adverse event for a patient. The same survey found that patients who receive denial letters frequently do not understand that they have the right to appeal, do not know how to navigate the appeals process, and often give up when faced with paperwork requirements and processing times that extend treatment delays further. The insurance denial system is, in many ways, designed around the premise that most people will stop fighting. Source: American Medical Association, "2022 AMA Prior Authorization Physician Survey," February 2023.

Your Rights in 2026: What the New Rules Actually Require

The regulatory environment has shifted meaningfully since the 2022–2023 investigations. The changes are significant but incomplete — and understanding what the rules now require is part of knowing your rights when you receive a denial.

• CMS Prior Authorization Rule (effective January 2024): The Centers for Medicare & Medicaid Services finalized rules in November 2023 requiring Medicare Advantage plans to use clinical criteria consistent with traditional Medicare when making coverage determinations. The rules explicitly prohibit Medicare Advantage plans from using 'internal coverage criteria or algorithms' to deny care that traditional Medicare would cover. These rules also set binding time limits: prior authorization decisions for non-urgent requests must be made within 72 hours; urgent decisions must be made within 24 hours. Plans that violate these timelines are subject to CMS enforcement action. Source: CMS Medicare Advantage Prior Authorization Final Rule, November 2023, effective January 2024.
• CMS Interoperability and Prior Authorization Rule (effective January 2026): A second rule finalized by CMS requires Medicare Advantage plans, Medicaid managed care plans, and ACA marketplace plans to implement electronic prior authorization systems and provide standardized, machine-readable data on prior authorization processing — including denial rates and approval times. This transparency requirement is designed to make systematic denial patterns visible to regulators and patients in a way they were not before. Source: CMS Interoperability and Prior Authorization Final Rule, 2024.
• State legislation: Multiple states have moved to restrict AI-only denial systems since 2023. California's prior authorization reform laws, Colorado's SB 23-063 (requiring insurance carriers to process prior authorizations on a real-time basis), and similar legislation in Illinois and New York have established a patchwork of state-level protections that vary significantly by location. The general trend is toward requiring that adverse coverage determinations be reviewed by a licensed clinician in the same specialty as the treating physician, not an algorithm or an out-of-specialty reviewer. Source: National Conference of State Legislatures, 2025; California Department of Managed Health Care, 2024.
• Ongoing litigation: As of April 2026, class action lawsuits related to UnitedHealthcare's nH Predict algorithm and similar claims against other major insurers are in various stages of litigation. No final judgments had been entered in the major class action cases as of publication. The litigation itself, regardless of outcome, has created additional documentation of how these systems operate and increased regulatory scrutiny. Source: federal court dockets, 2024–2026.

The Six-Step Fight-Back Guide: How to Appeal and Win

The following steps apply to the most common scenario: you or your provider has received a prior authorization denial or a post-service claim denial from your health insurer. The process differs in some details between Medicare Advantage, ACA marketplace plans, Medicaid, and employer-sponsored plans — but the general architecture is the same. Each step below includes the specific action to take, why it matters, and what to say.

• Step 1 — Get the denial in writing and identify the specific reason. Every insurer is required to provide a written explanation for a denial, including the specific clinical criteria or coverage policy cited. Request this immediately if you have not already received it. The written denial is your roadmap: it tells you exactly which criterion the insurer claims you failed to meet, and therefore exactly what evidence you need to provide on appeal. Many denials cite a vague 'not medically necessary' determination — your first task is to obtain the specific guideline or criteria document the insurer used. You are legally entitled to a copy of the specific utilization management guidelines cited. Request them explicitly by name in writing. Sources: ACA Sec. 2719; ERISA regulations for employer-sponsored plans; Medicare Advantage regulations (42 CFR Part 422).
• Step 2 — Request a peer-to-peer review immediately. A peer-to-peer review is a phone call between your treating physician and the insurer's reviewing physician (the one who signed off on the denial). This step is not always prominently disclosed in denial letters, but it is available in the vast majority of insurance plans and is one of the most effective early-stage interventions. The logic is simple: an insurer's reviewing physician is less likely to maintain a denial when challenged directly by the specialist who has examined the patient and can articulate the specific clinical circumstances the algorithm did not capture. Physicians who request peer-to-peer reviews succeed in reversing denials at a high rate according to AMA survey data. Your doctor's office should request this; if they have not, ask them to. The window for peer-to-peer reviews is typically short — often 24 to 72 hours after the denial. Do not wait. Source: AMA advocacy on prior authorization reform, 2022–2025.
• Step 3 — File your internal appeal with complete clinical documentation. If the peer-to-peer review does not resolve the denial, file your formal internal appeal immediately. Internal appeals are the first official step in the legal appeal process and have strict deadlines — typically 180 days for ACA marketplace plans, 60 days for Medicare, and varying periods for employer-sponsored plans. Your appeal letter should include: (a) your physician's letter of medical necessity with specific clinical detail — not a form letter, a personalized letter that addresses the specific denial reason and explains why the standard criteria do not apply to your individual case; (b) any peer-reviewed literature supporting the medical necessity of the treatment; (c) any prior treatment records demonstrating that less intensive alternatives have been tried and failed (if the insurer is denying on 'step therapy' grounds); (d) explicit citation of the new CMS rule (if you have Medicare Advantage) prohibiting algorithm-only denials inconsistent with traditional Medicare coverage; and (e) a statement that your physician has determined the treatment is medically necessary and that you are requesting a clinical review by a board-certified physician in the relevant specialty. Source: CMS Medicare Appeals process; HHS ACA appeal regulations.
• Step 4 — If the internal appeal is denied, request an expedited external review. External review is an independent process conducted by an Independent Review Organization (IRO) that is not affiliated with your insurer. Under the ACA, you are entitled to external review after exhausting internal appeals for non-grandfathered plans. For Medicare Advantage, external review is conducted through the Medicare appeals process (ALJ hearing, Medicare Appeals Council, federal court). KFF Health News data shows that external reviewers overturn insurer decisions in a meaningful percentage of cases — in some studies, approximately 40 percent of external review decisions favor the patient. The key is that external reviewers are independent clinicians, not the insurer's utilization management team or algorithm. Source: KFF Health News, 2023; HHS external review regulations; CMS Medicare Appeals process.
• Step 5 — File a complaint with your state insurance commissioner in parallel. Filing a formal complaint with your state's insurance regulator while your appeal is pending does two things: it creates an official record that will be tracked in the insurer's compliance file, and it triggers a regulatory inquiry that the insurer must respond to. In states with strong insurance departments — California, New York, Colorado — this step has meaningful teeth. If the denial appears to violate the new CMS prior authorization rules (for Medicare Advantage) or your state's prior authorization laws, cite those specific violations in your complaint. State insurance departments are required to investigate formal complaints and respond to the complainant. Source: National Association of Insurance Commissioners (NAIC); state insurance department resources.
• Step 6 — Contact a patient advocate or attorney if the claim is significant. For claims involving major medical costs — hospitalizations, surgeries, long-term care, cancer treatment, specialty drugs — the financial stakes justify professional help. Patient advocates are specialists who navigate insurance appeals on behalf of patients; many are nurses or physicians who understand both the clinical and administrative dimensions. Patient advocacy services range from non-profit organizations (the Patient Advocate Foundation provides free case management services to patients with chronic illness and life-threatening diseases) to for-hire advocates who charge a percentage of recovered benefits. For clear coverage violations — particularly cases where a Medicare Advantage plan violated the new CMS rules on algorithmic denials — healthcare attorneys who specialize in insurance disputes can evaluate whether legal action is appropriate. The cost of a consultation is often justified by the size of the claim at stake. Source: Patient Advocate Foundation (patientadvocate.org); National Patient Advocate Foundation.

If You Have Medicare Advantage: Your Specific Rights Under the 2024 Rules

Medicare Advantage enrollees have a specific and powerful additional protection that took effect in January 2024. CMS's final rule explicitly prohibits Medicare Advantage plans from using 'internal coverage criteria or algorithms' as the basis for denying coverage of services that traditional Medicare covers. This rule was a direct response to the Senate Finance Committee's findings and the pattern of algorithmic denials documented in the 2022–2023 investigations. If you are a Medicare Advantage enrollee and your plan has denied a prior authorization or coverage decision for a service that traditional Medicare would cover, that denial may be a specific regulatory violation — not merely a coverage dispute. Source: CMS Medicare Advantage Prior Authorization Final Rule, November 2023.

In practical terms, this means: if you receive a denial citing 'not medically necessary' and the service in question is one that traditional Medicare would cover (you can verify this at medicare.gov or by calling 1-800-MEDICARE), you should cite this rule explicitly in your internal appeal and in your complaint to your state insurance department. You should also file a complaint directly with CMS at 1-800-MEDICARE. CMS tracks Medicare Advantage compliance with prior authorization requirements and can conduct audits of plans with elevated denial rates. The combination of a state insurance complaint and a CMS complaint simultaneously creates the most effective regulatory pressure on the plan while your formal appeal is pending. Source: CMS Medicare Advantage compliance resources, 2024.

What the Algorithm Cannot See — And Why That's Your Strongest Argument on Appeal

The fundamental vulnerability of an AI denial system — the thing that makes appeals winnable at the rates the data shows — is that an algorithm trained on population averages cannot evaluate your specific clinical situation. Insurance AI denial tools make decisions based on diagnostic codes, procedure codes, and statistical population data. They do not have access to your physician's clinical reasoning, your particular anatomy, your response to prior treatments, the nuances of your symptoms, or the professional judgment of the specialist who has examined you directly. Every successful appeal rests on exactly this gap: the difference between what the algorithm decided based on aggregate data and what your treating physician determined based on direct clinical evaluation of you as an individual.

This is why the physician letter of medical necessity — step 3's most important component — is not a formality. It is the primary evidence that an algorithmic denial system structurally cannot produce and cannot refute. A well-written letter of medical necessity from your treating physician that specifically addresses why the standard population-level criteria do not apply to your individual case is the most effective instrument in a health insurance appeal. It should not be a form letter or a brief note. It should document: your specific diagnosis and symptoms, the treatments already tried and why they failed or are contraindicated for you, the clinical basis for the recommended treatment, relevant peer-reviewed literature, and an explicit statement that the physician has reviewed the denial reason and explains in clinical terms why it does not apply to this patient. The more specific it is to you as an individual — rather than a generic argument that the treatment is effective in general — the harder it is to deny on appeal. Source: Patient Advocate Foundation; AMA appeal guidance resources.

What Comes Next: The AI Denial System Is Getting Regulated — But You Still Have to Fight

The use of AI in insurance utilization management is not going to reverse. The economics are too compelling, the technology is improving too rapidly, and the regulatory framework — while tightening — is not moving toward prohibition. What is changing is the accountability structure. The CMS rules that took effect in 2024 and 2026 create enforceable standards around what these systems can do and how quickly they must respond. State laws are creating physician-review requirements and transparency mandates. Class action litigation is creating financial incentives to reduce error rates. The direction of travel is toward AI-assisted utilization management with meaningful human clinical review, not toward a system where an algorithm's output is the final word on your care.

What that trajectory means for you right now, in April 2026, is this: the system is in transition. The new rules exist but are not uniformly enforced. The insurers subject to class action litigation are adjusting their processes but have not uniformly abandoned the algorithmic review frameworks. Your rights are clearer than they were in 2022, and your ability to enforce those rights through the appeal system is better supported by the new CMS timelines and external review requirements. But you have to use those rights. The appeal process works for people who navigate it. It does not work for the overwhelming majority of Americans who receive a denial letter, assume it is final, and never respond.

Frequently Asked Questions