Who decides what is healthy? Algorithmic classification and medical normativity

The proliferation of algorithmic classification systems in healthcare raises fundamental questions about medical normativity: who determines the standards of health, and through what mechanisms are those standards constructed and enforced? This paper makes three original contributions. First, it conceptualizes algorithmic normativity as a multidimensional phenomenon operating through training data selection, loss function design, and clinical deployment contexts. Second, it forges a novel analytical link between population-level structural inequity and the undertheorised phenomenological dimension of diagnostic identity fragmentation. Third, it proposes a human-centric governance framework integrating transparency, plural normativities, and meaningful patient participation as a coherent alternative to purely technical approaches to algorithmic fairness. Drawing on medical sociology, science and technology studies, and bioethics, I argue that algorithmic systems do not simply discover pre-existing medical truths but actively construct normative frameworks that reflect and reproduce social inequities embedded in their development. The question of who decides what is healthy cannot be answered through computational means alone but requires sustained democratic engagement with the ethical, social, and political dimensions of algorithmic medicine.