CRA was retained on behalf of a practice management software company facing False Claims Act allegations that its software was biased to encourage provision of healthcare services. Relator proposed a statistical sampling methodology to demonstrate liability in a bellwether trial, but CRA analysis demonstrated that the proposed methodology itself was significantly biased. In addition, CRA refuted expert reports from a statistician (proposing alternate sampling methods), a reimbursement specialist (regarding billing and reimbursement mechanisms and whether they could be affected by the alleged behavior), and a medical coding expert (regarding the link between treatment recommendations and actual billed services). The case resolved with summary judgement against the Relator.
Confirmation bias in LLM pricing recommendations
In a new article, CRA’s Maxime Cohen and co-author Eddy Hage-Youssef test this assumption through a controlled pricing experiment using Claude Haiku 4.5 and...