MedPath: Multi-Domain Cross-Vocabulary Hierarchical Paths for Biomedical Entity Linking

MedPath is a newly released large-scale multi-domain biomedical entity linking dataset that integrates nine expert-annotated datasets, normalizes entities using the Unified Medical Language System (UMLS), and includes mappings to 62 biomedical vocabularies along with enriched ontological paths across 11 vocabularies. This dataset addresses challenges in biomedical Named Entity Recognition and Linking by providing a resource for training and evaluating semantic-rich and interpretable models. Its development is significant for practitioners aiming to create interoperable and explainable clinical NLP systems, advancing the field of biomedical natural language processing.