The Decision Context Record™: A Framework for Demonstrating Effective Human Oversight in Healthcare AI

Healthcare AI organizations operating under the European Union AI Act and the Medical Device Regulation face an increasingly precise evidential obligation. Article 14 of the EU AI Act requires not merely that human oversight exist, but that it be capable of being exercised effectively. General Safety and Performance Requirement (GSPR) 14.2(d) of the MDR further demands that devices perform as intended under foreseeable conditions, an obligation that extends to the conditions under which AI-assisted clinical decisions occur. This paper argues that the predominant approach to human oversight in healthcare AI, including approval logging, governance documentation, and audit trails, addresses process intent but fails to constitute operational evidence of effective oversight. We introduce the Decision Context Record™ (DCR), a structured evidentiary framework comprising six component categories: AI Output Context, Human Context, Decision Environment, Intervention Context, Decision Outcome, and Interaction Evidence. We further introduce the Decision Reconstruction Test™, a five-question evaluative instrument for assessing whether oversight can be independently verified after a decision has occurred, and the Operational Oversight Maturity Model™, describing five levels of organisational readiness. The central thesis of this paper is that human approval is not automatically human oversight, and that oversight cannot be demonstrated without evidence. The DCR represents a practical operational response to the regulatory evidence gap that exists between governance intention and verifiable oversight.