Sarah Graham

This article investigates ethical hazards associated with argumentative shifts following the emergence of combined non-inferiority and efficiency/revenue facilitation (NERF) endpoints, with particular attention to research on artificial intelligence (AI) in health and medicine. The study presented here adopts a Toulmin argumentative analysis approach to investigate the dominant persuasive logics of twenty-three health AI trials evaluating NERF endpoints. In so doing, the article demonstrates how the argumentative logics of NERF trial reports shifts away from health outcomes as the putative evidence base for adopting novel interventions toward an argumentative model that prioritizes economic benefits over patient benefits. Ultimately, the article argues that new logic of NERF endpoints is consistent with and risks accelerating the harms to patient care that arise from increased financialization of the healthcare sector. If the use of NERF endpoints and attendant logics continue to grow unchecked, they could result in significant harms to patient health and well-being.