BioNNE-L Shared Task at BioASQ 2025

Shared Task Overview

The BioNNE-L shared task challenges participants to tackle medical concept normalization (MCN), also known as entity linking, for English and Russian languages.

Goal: map biomedical entity mentions to their corresponding concept names and unique identifiers (CUIs) within the Unified Medical Language System (UMLS).

Data: Entities from English and Russian scientific abstracts in the biomedical domain. The BioNNE-L task utilizes the MCN annotation of the NEREL-BIO dataset [1], which provides annotated mentions of disorders, anatomical structures, chemicals, diagnostic procedures, and biological functions.

Evaluation Tracks: Similar to the BioNNE 2024 task [2], the evaluation is structured into three tracks:

• (1), (2) Monolingual tracks requiring separate models for English and Russian;
• (3) Bilingual track: requiring a single model trained on multilingual dataset combined from English and Russian data:

Shared Task-Specific Challenges:

• Nestedness: Complexity of nested entity mentions;

• Partial terminology: a concept does not have concept name in low-resource language (Russian) and thus has to be linked to a vocabulary entry in rich-resource language (English).

Data

The dataset entities can be loaded from HuggingFace:

dataset = load_dataset("andorei/BioNNE-L", "Bilingual", split="train")

Annotated Data Format

Each line describes a single biomedical entity of possible entity types: (i) Disease (DISO), (ii) Chemical (CHEM), (iii) Anatomy (ANATOMY).

• doc_id is a unique textual document identifier the given entity is derived from. Each document contains multiple entities described with their spans in the document;

• text is a textual mention string of the given entity;

• entity_type can take one of three values: DISO, CHEM, ANATOMY. These are high-level semantic types supported by the underlying UMLS knowledge base;

• spans provides a list of comma-separated entity positions within the given textual document with id doc_id. Each span entry provides starting and ending positions, e.g., 22-28. An entity provided with multiple positions (e.g., 472-476,492-500 for lung injuries) corresponds to an interrupted entity with non-entity words inserted between entity words;

• UMLS_CUI is the Concept Unique Identifier (CUI) in the UMLS metathesaurus (UMLS serves the normalization vocabulary). This field provides ground truth CUI for the given entity. We note that the predicted CUI in your submission file must be in prediction column.

Normalization Vocabulary

In our work, we collect the bilingual concept vocabulary derived from English and Russian UMLS parts. Due to incompleteness of Russian vocabulary (Partial terminology challenge), part of Russian entities have to be mapped to an English vocabular entry. Vocabulary file is a tsv file with the following fields:

CUI - UMLS CUI;

semantic_type - Concept's semantic type (DISO/CHEM/ANATOMY);

concept_name is a textual concept name derived from UMLS. Each concept can have multiple vocabular entries with different names but sharing the same CUI.

References

[1] Loukachevitch, Natalia, Andrey Sakhovskiy, and Elena Tutubalina. "Biomedical Concept Normalization over Nested Entities with Partial UMLS Terminology in Russian." Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024). 2024.

[2] Davydova, Vera, Natalia Loukachevitch, and Elena Tutubalina. "Overview of BioNNE task on biomedical nested named entity recognition at BioASQ 2024." CLEF Working Notes (2024).