Medicare Advantage plans are Medicare-approved insurance plans run by private health companies, such as UnitedHealthcare, Humana, and regional health plans. These plans give options besides the usual federal Medicare and often include extra benefits. The benefits process includes reviewing and approving many claims for services like rehabilitation and care after hospital stays.
Recently, insurers have started using AI models to make claim decisions faster. Algorithms look at clinical data, doctors’ notes, and past claim results to say yes or no to coverage. One example is an AI tool called “nH Predict,” made by NaviHealth, which is part of UnitedHealth Group. Insurers use it to guess how long patients will stay in rehab or nursing facilities and decide if coverage should continue, be limited, or stopped.
Though AI could help, lawsuits and reports show some AI tools from big insurers have very high error rates. For example, the nH Predict tool has about a 90% error rate. This means 9 out of 10 AI decisions are wrong when checked through appeals or official reviews.
These errors have big effects on older patients in Medicare Advantage plans:
This problem is not only at UnitedHealth. Other big insurers like Humana and Cigna also use AI to help decide on claim denials. For example, Cigna’s PxDx algorithm denied over 300,000 claims in two months, sometimes without careful individual checks.
The U.S. Senate Permanent Subcommittee on Investigations reported that insurers are using AI more to cut costs by denying more claims. This raises ethical concerns about balancing cost savings with patient wellbeing. Lawsuits argue that insurers acted unfairly or broke contracts by using flawed AI.
Experts and organizations say it is important to keep a human involved in decisions influenced by AI. The American Medical Association (AMA) says AI can speed up approvals but states that human reviews must happen before care is denied.
People who manage healthcare practices should watch how AI affects patient care access, payments, and work flow when handling Medicare Advantage claims.
Automation in claims processing has both good and bad sides. AI can do 50% to 75% of manual insurance approval work, which speeds things up and lowers admin costs, helping patients and providers when used well.
But the case with UnitedHealth’s nH Predict shows that AI with many errors can hurt patient care if it ignores doctors or unique patient issues.
Good automated workflows need safeguards like:
Healthcare IT managers have a key job designing systems that use AI but do not give full control to tools that are not yet very accurate.
For administrators and owners, knowing that AI is a tool—not a replacement for human judgment—is important. They should push for systems that focus on patient care and use AI to cut delays, not skip necessary human checks.
The risks affect real people. Families like those of Gene Lokken and Dale Tetzloff show the cost behind the problem. Both men needed longer rehab after illness or injury, and doctors agreed on this. Still, AI denials by UnitedHealth made their families pay large amounts or risk worse health.
Data from lawsuits shows most AI denials are later reversed on appeal or federal review. But most patients don’t know they can appeal or can’t handle the complex process. They then skip care or face big bills.
Legal experts say the problem is not just AI technology but that insurers use it mainly to save money, knowing few people appeal. Attorney Ryan Clarkson says companies put profits first by using high-error AI, taking advantage of rare appeals.
Medical practice administrators and owners in the U.S. face the challenge of working with insurance systems driven more by AI. They should focus on:
Good preparation means balancing tech use with ethical care. This helps make sure AI improves rather than harms patient care.
This review of AI in Medicare Advantage claims shows the results of relying on flawed algorithms instead of patient care. With careful management, teamwork, and smart tech use, medical administrators and health IT leaders can help lower risks, support fair coverage choices, and protect patients who depend on Medicare Advantage benefits.
Families of two deceased former beneficiaries filed a lawsuit claiming UnitedHealth used a faulty AI algorithm to deny necessary Medicare coverage, resulting in financial and medical hardships for elderly patients.
The AI model, known as ‘nH Predict,’ reportedly has a 90% error rate according to the lawsuit.
These are Medicare-approved insurance plans administered by private insurers like UnitedHealth, providing alternatives to traditional federal Medicare coverage.
The lawsuit claims it led to premature denial of coverage for care deemed necessary by physicians, forcing patients into tough financial situations.
NaviHealth states that the AI tool is used as a guide to help inform providers on patient care needs, not for making coverage decisions.
The lawsuit mentions that roughly 0.2% of policyholders appeal denied claims, with most either paying out-of-pocket or forgoing care.
McKinsey reported that AI could automate 50%-75% of manual tasks involved in insurance approvals, potentially leading to faster turnaround times.
The AMA appreciates AI’s potential but advises that insurers should ensure human review of patient records before denying care.
A ProPublica review revealed that Cigna doctors rejected over 300,000 claims within a two-month period using artificial intelligence.
The lawsuit may represent broader concerns about AI’s reliability in healthcare and its implications for patient rights and care efficacy.
Simbo is using AI agents to automate all phone related workflows. Connect now to know more.
Schedule AI Agents Demo Now© Since 2019, Simbo AI.
