A machine-assisted framework for systematic error analysis in clinical concept extraction
Error analysis is critical for evaluating and improving clinical natural language processing (NLP) models, yet no standardized framework exists to support systematic error analysis for clinical concept extraction. Here we present MedError, a machine-assisted, human-in-the-loop framework integrating a validated error ta...