Workshop at ICML 2026 exploring the algorithmic frontier of post-deployment improvement for medical AI. As AI models are deployed in clinical settings worldwide, this workshop addresses the gap between strong benchmark performance and real-world clinical reliability — focusing on deployment-time diagnostics, human expert feedback, and model improvement under the stringent constraints of healthcare (high-risk decisions, noisy feedback, heterogeneous populations, and safety/equity requirements). While grounded in healthcare, the algorithms discussed are broadly relevant to deploying models in any high-stakes environment.
pnpm installpnpm devThe site will be available at http://localhost:3000.
pnpm buildpnpm lint