Los Angeles attorney Ryan Clarkson says some of the country’s biggest health insurers are quietly leaning on artificial intelligence to deny medically recommended care in bulk, leaving seniors and other patients stuck with surprise medical bills. His firm has filed a string of federal class actions that claim automated systems are effectively making coverage calls with little real human review, even though internal appeals often reverse those denials. Clarkson raised the alarm in a TV interview this week and said the lawsuits aim to pry open how insurers are using predictive models to manage care.
What the lawsuits allege
Plaintiffs say insurers have rolled out a set of automated tools, including NaviHealth’s nH Predict, used by some Medicare Advantage administrators and Cigna’s PxDx review system, to set “target” lengths of stay and reject claims en masse. The suits argue that those targets became de facto decisions instead of guidance, and families of patients say the rigid cutoffs disrupted rehabilitation, triggered hefty out-of-pocket bills, and in some cases coincided with worse health outcomes, according to ProPublica.
Evidence in the record
Court complaints filed by Clarkson’s firm and others describe examples where coverage stopped on the exact date generated by an algorithm and say many of those denials were later overturned on appeal, which plaintiffs argue is a red flag for how unreliable the tools can be, as described in the court complaint. The filings also claim staff were discouraged from second-guessing model predictions and that key decision information was treated as proprietary instead of being shared with treating physicians. That secrecy is framed as a central threat to patient care, according to Courthouse News.
Insurers push back
Insurers insist they are not letting software overrule doctors. They say the systems are simply tools to help guide decisions, not replace clinical judgment. An Optum Health spokesperson told Medical Economics that the naviHealth tool "is not used to make coverage determinations," and Clarkson repeated his concerns in an interview with CBS Los Angeles. Cigna labeled the PxDx allegations "highly questionable" in a statement to CBS News.
Why appeals and oversight matter
Plaintiffs say only a relatively small share of members actually contest denials, so even if appeals often succeed, most people are still stuck with the insurer’s first answer. Reporting and court filings cite internal data and appeal outcomes to argue that automated denials are frequently reversed at internal or administrative review, a pattern critics say reinforces calls for independent oversight and far clearer disclosure about how these tools are used, according to STAT and the court filings.
Where the cases stand
Several of the lawsuits are now moving forward in federal court, and judges in some cases have rejected insurers’ attempts to limit discovery, a step plaintiffs hope will force disclosure of internal records on how the algorithms were built and rolled out, according to Becker's. Other courts have allowed breach-of-contract and bad-faith claims to proceed in related cases, signaling that judges are at least willing to test whether insurers can lean on opaque models instead of individualized medical review, as reported by Ars Technica.
What this could mean
Legal experts say the litigation could set a key precedent on whether insurers are allowed to outsource clinical judgments to black-box algorithms or whether state and federal rules require documented human review for every claim, a question regulators and researchers are increasingly probing, according to a JAMA Health Forum analysis. For now, the court fights are shaping up as a real-time test of industry practices, regulatory muscle, and just how transparent insurers will have to be about the automated systems that help decide who gets care and who gets the bill.
