Abstract Diagnostic rules are codified in consensus manuals such as DSM-5, yet they remain written in narrative form and cannot be computationally interrogated. Here, a deterministic framework is presented that translates diagnostic criteria into a machine-actionable representation of the full symptom space, which can be charted, navigated, and systematically analyzed. Unlike probabilistic models that infer patterns from large textual corpora, this framework directly interrogates explicit consensus criteria, providing a transparent and reproducible means of assessing conceptual coherence. Its potential is demonstrated by charting schizophrenia-spectrum disorders, which remain conceptually distinct despite substantial symptom overlap, and by evaluating the current National Academies’ definition of Long COVID, which is largely subsumed by depressive and anxiety disorders. By making diagnostic consensus computable, the framework provides a reproducible foundation for evaluating delineation properties of existing and candidate diagnostic constructs and for developing interpretable, regulatory-compliant diagnostic support tools.
Strasser-Kirchweger et al. (Mon,) studied this question.