Annotating and indexing scientific articles with rare diseases

Area of Science:

• Biomedical Informatics
• Computational Biology
• Rare Disease Research

Background:

• Approximately 30 million Europeans have rare (orphan) diseases, affecting fewer than 1 in 2,000 individuals.
• Identifying scientific literature and guidelines for specific rare diseases presents a significant challenge.
• Existing methods are hindered by limited annotated data and variations in rare disease name representation.

Purpose of the Study:

• To develop a novel methodology for annotating and indexing scientific text with rare disease concepts from the OrphaNet taxonomy.
• To address challenges of data scarcity and lexical variability in rare disease literature.
• To enable scalable, automated identification of rare disease research.

Main Methods:

• A framework was developed integrating the TERMite engine with OrphaNet.
• Key components include curated synonym expansion, label normalization (handling deprecated/renamed concepts), and fuzzy matching.
• The pipeline was applied to Scopus to create disease-specific corpora for bibliometric and scientometric analyses.

Main Results:

• The approach achieved 92% precision, 75% recall, and 83% F1 score on benchmark datasets, outperforming a string-matching baseline.
• The system generates disease-specific corpora suitable for analyzing research activity (e.g., by institution, country, subject area).
• Outputs power the Rare Diseases Monitor dashboard for exploring research trends.

Conclusions:

• This work presents the first systematic, scalable semantic framework for annotating and indexing rare disease literature.
• The automated, reproducible pipeline advances biomedical semantics for rare diseases.
• The framework enables disease-centric monitoring, evaluation, and discovery within the research landscape.