UnitedHealthcare's nH Predict algorithm increased skilled nursing facility denial rates ninefold between 2020 and 2022 for Medicare Advantage enrollees

UnitedHealthcare used an AI tool called nH Predict, developed by its subsidiary NaviHealth, to determine how long elderly Medicare Advantage patients should receive post-acute care in skilled nursing facilities. A federal lawsuit alleges the algorithm has a known 90% error rate, yet UnitedHealthcare used it to override treating physicians' clinical judgments and prematurely terminate coverage for elderly patients still in active recovery. Why it matters: Elderly patients recovering from hip replacements, strokes, and other serious conditions are discharged from skilled nursing facilities before they can safely care for themselves, so they suffer falls, infections, and medical setbacks at home without professional monitoring, so they end up readmitted to hospitals at higher acuity levels, so Medicare bears greater costs than the original skilled nursing stay would have required, so the financial savings UnitedHealthcare captures from early discharges are externalized as higher costs to the broader Medicare system and devastating health outcomes for vulnerable seniors. The structural root cause is that Medicare Advantage insurers receive a fixed per-member-per-month capitation payment from CMS, creating a direct financial incentive to minimize post-acute care spending. Unlike traditional Medicare where providers bill fee-for-service, MA insurers profit by spending less than their capitation payment, which structurally aligns their financial interests against providing the duration of care that physicians recommend.