AAbAAC: An Annotated Corpus for Autoimmunity Information Extraction

2026-06-25 · Source: Paper Index on ACL Anthology · Field: Science & Research — Health & Medical Research, Mathematics & Computational Sciences · Depth: Expert, medium

Summary

Fabien Maury, Solène Grosdidier, Maud De Dieuleveult, and Adrien Coulet present AAbAAC (AutoAntibodies and Autoimmunity Annotated Corpus), a new resource designed to enhance information extraction in the specialized biomedical field of autoimmunity. This corpus comprises 115 abstracts manually selected from PubMed and annotated for key entities such as autoimmune diseases, autoantibodies, their molecular targets, body locations, and associated clinical signs. Developed to address performance limitations of generalist deep learning and large language models in highly complex domains, AAbAAC was utilized to evaluate and fine-tune named entity recognition (NER) models. The study, presented at BioNLP 2026, confirmed AAbAAC's utility, demonstrating improved NER performance after fine-tuning, thereby underscoring the effectiveness of targeted, small-scale annotation efforts for specialized scientific domains. The corpus is publicly available on GitHub.

Key takeaway

For NLP Engineers or Research Scientists working on biomedical information extraction, you should consider developing or utilizing specialized annotated corpora like AAbAAC. If your generalist models underperform in niche domains such as autoimmunity, fine-tuning them with a targeted dataset can yield significant performance improvements. This approach validates investing in small-scale annotation efforts to overcome the complexity challenges inherent in highly specialized scientific texts, directly enhancing the accuracy of your named entity recognition systems.

Key insights

Small, specialized annotated corpora significantly improve information extraction in complex biomedical domains.

Principles

• Generalist models struggle in highly specialized fields.
• Domain-specific annotation enhances NER performance.
• Small-scale efforts yield valuable domain resources.

Method

AAbAAC was created by manually annotating 115 PubMed abstracts for autoimmune diseases, autoantibodies, molecular targets, body locations, and clinical signs, then used to evaluate and fine-tune NER models.

In practice

• Develop targeted corpora for niche biomedical NLP tasks.
• Fine-tune existing NER models with domain-specific data.
• Utilize AAbAAC for autoimmunity research.

Topics

• Autoimmunity
• Information Extraction
• Named Entity Recognition
• Annotated Corpus
• Biomedical NLP
• Deep Learning Models

Code references

• f-maury/AAbAAC

Best for: AI Scientist, NLP Engineer, Research Scientist

Related on AIssential

• AAbAAC: An Annotated Corpus for Autoimmunity Information Extraction — Small, specialized annotated corpora significantly boost information extraction performance in niche biomedical domains.
• AI-assisted Protocol Information Extraction For Improved Accuracy and Efficiency in Clinical Trial Workflows — AI-assisted RAG significantly boosts clinical trial protocol data extraction accuracy and efficiency over standalone LLMs.
• Scaling antibody language models improves structure aware representation for antibody engineering — Scaling antibody language models with advanced tokenization improves structure-aware representation for engineering.
• Predicting relations between SOAP note sections: The value of incorporating a clinical information model — Fine-tuning a RoBERTa-base model on manually annotated clinical data improves NER for SOAP note sections.
• RacismoBR: A Manually Annotated Dataset for Racist Discourse Detection in Brazilian Portuguese — Culturally grounded annotation is more critical than model architecture for effective racism detection.
• Nested Named Entity Recognition in Plasma Physics Research Articles — A lightweight BERT-CRF model with entity-specific specialization and Bayesian Optimization effectively extracts nested entities from plasma physics texts.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.