AI RESEARCH
End-to-End Evaluation and Governance of an EHR-Embedded AI Agent for Clinicians
arXiv CS.AI
•
ArXi:2604.27309v1 Announce Type: new Clinical AI systems require not just point-in-time evaluation but continuous governance: the ongoing practice of monitoring, evaluating, iterating, and re-evaluating performance throughout deployment. We present an end-to-end framework of governance that integrates rubric validation, live deployment feedback, technical performance monitoring, and cost tracking, with controlled experimentation gating system changes before deployment.