Tag
This paper addresses the problem of treatment-induced label indeterminacy in clinical prediction, using post-cardiac-arrest neurological prognostication as a case study. The authors propose a framework that incorporates expert annotations of counterfactual outcomes and highlight tradeoffs between accuracy on certain versus uncertain cases.
UC Berkeley researchers trained an AI model on hundreds of thousands of EKGs to detect a previously unrecognized signal that predicts sudden cardiac death risk more accurately than current methods, potentially saving thousands of lives annually.