Regulatory evidence convergence for medical device artificial intelligence

Article-only preprint version of a regulatory-science manuscript proposing a residual-based lifecycle evidence-convergence framework for medical device artificial intelligence. This version is provided for scholarly timestamping and discussion only. It does not include software, scoring workbooks, implementation code, calibration models, commercial templates, client-facing assessment tools, or operational deployment materials. The manuscript has been submitted to PLOS Digital Health as manuscript PDIG-D-26-01055 and has not yet been peer reviewed. No regulatory approval, certification, notified-body acceptance, conformity-assessment validity, legal compliance conclusion, or operational readiness threshold is claimed. Commercial software, tooling, training, audit templates, implementation packages, and service workflows related to this framework are not included in this record and may be subject to separate intellectual-property, licensing, trademark, copyright, or trade-secret protection.