nH Predict algorithm accused of having a 90 percent error rate

UnitedHealth Group is accused of using an AI-powered algorithm to systemically deny post-acute healthcare claims for Medicare Advantage patients, raising concerns about its accuracy and reliability, and about UnitedHealth Group's governance. What happened UnitedHealth Group's nH Predict system resulted in the wrongful denial of claims for medically necessary extended care for elderly and disabled patients under Medicare Advantage plans 90 percent of the time, a ccording to a lawsuit . Filed on behalf of the families of two deceased patients - 91-year-old Wisconsin resident Gene B. Lokken and 74-year-old Dale Tetzloff - who had UnitedHealthcare coverage, the suit says the alleged error rate was based on the percentage of patient claims that the company initially denies and then gets "reversed through either an internal appeal process or through federal Administrative Law Judge (ALJ) proceedings." The denials had severe consequences, including unfa ir forced early discharge from care facilities, deteriorating patient health conditions ultimately r esulting in their premature deaths , and signi ficant financial burdens , the suit said. Lokken's family had to pay roughly USD 12,000 to USD 14,000 per month out-of-pocket for about a year of therapy until his death. UnitedHealth Group's use of nH Predict was found by a 2024 US Senate investigation into denial rates for coverage through Medicare Advantage to coincide with a massive increase in claim denials . D enials for post-acute care reportedly increas ed by 227 percent in 2022 alone. Why it happened UnitedHealth Group stands accused of pressuring managers to adhere to the algorithm's recommendations, even against the advice of clinicians and families, in order to save time and money. The use of these kinds of algorithms is estimated to automate 50-75 percent of human labour involved in claims processing. Lawyers for UnitedHealth Group argue that the lawsuit should be dropped because plaintiffs failed to complete Medicare’s appeals process and that their grievances are with the federal government and not the company. What it means nH Predict's high error rate raises concerns about its accuracy and reliability, and about UnitedHealth Group's governance, notably the its transparency and accountability of its automated systems, and its values and ethics. Rumours circulated widely about nH Predict after the fatal shooting of UnitedHealthcare CEO Brian Thompson by Luigi Mangione in December 2024. Thompson had previously faced criticism for the company's rejection of insurance claims. The case also highlights the risks of inadequate transparency, oversight and accountability of AI in health insurance, and is seen as likely to undermine trust in automated healthcare decision systems generally. System 🤖 nH Predict Developer: Optum; UnitedHealth Group Country: USA Sector: Health Purpose: Predict post-acute care needs Technology: Prediction algorithm; Machine learning Issue: Accountability; Accuracy/reliability; Transparency Legal, regulatory 👩🏼‍⚖️ The Estate of Gene B. Lokken and The Estate of Dale Henry Tetzloff, individually and on behalf of all others similarly situated, vs. UNITEDHEALTH GROUP, INC., UNITEDHEALTHCARE, INC., NAVIHEALTH, INC., and DOES 1-50 Fact check 🚩 Snopes. The Facts Behind Accusations That UnitedHealthcare Uses Error-Prone AI to Deny Claims

Date it happened

2023-11-01

Product, system or model

nH Predict

This is a record of what was reported, not a finding that anyone broke the law. If it names your organisation and you believe it is wrong, the corrections process is free and open to everyone.