Human-in-the-Loop
A design principle where a clinician reviews and can override an AI system's output before it affects care.
Human-in-the-loop describes an AI system design in which a clinician reviews, and can accept, edit, or override, the AI's output before it affects patient care, rather than the system acting fully autonomously.
Most current clinical AI, including ambient documentation and diagnostic triage tools, is deployed as human-in-the-loop. Fully autonomous clinical AI, where no clinician review is required before an output is acted on, remains a narrow, more tightly regulated category.