AI-enabled medical devices increasingly depend on software updates, model evolution, and continuous security maintenance, yet conventional DevOps does not satisfy the traceability, accountability, and governance obligations required for safety-critical regulation. This paper introduces BioDevOps, a framework-in-progress for continuous validation and deployment in regulated biomedical systems. The current contribution combines a four-layer UML architecture, a compact formalization of governance invariants, and a small deployment scenario illustrating how runtime risk signals are transformed into governance evidence before release decisions are made. It also identifies the minimal semantic ingredients required for later verification-oriented encoding. More specifically, the paper makes explicit the separation between automated evidence production and human-controlled release authorization, and frames this separation as the key condition for reconciling adaptive software evolution with accountable regulatory oversight.
PhD Forum: BioDevOps: Toward a Formal Framework for Continuous Deployment in Safety-Critical Medical Devices
Fabio Orazio Mirto
2026-01-01
Abstract
AI-enabled medical devices increasingly depend on software updates, model evolution, and continuous security maintenance, yet conventional DevOps does not satisfy the traceability, accountability, and governance obligations required for safety-critical regulation. This paper introduces BioDevOps, a framework-in-progress for continuous validation and deployment in regulated biomedical systems. The current contribution combines a four-layer UML architecture, a compact formalization of governance invariants, and a small deployment scenario illustrating how runtime risk signals are transformed into governance evidence before release decisions are made. It also identifies the minimal semantic ingredients required for later verification-oriented encoding. More specifically, the paper makes explicit the separation between automated evidence production and human-controlled release authorization, and frames this separation as the key condition for reconciling adaptive software evolution with accountable regulatory oversight.Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.


