Precision information extraction for rare disease epidemiology at scale.

Abstract in brief

• The study addresses the challenge of collecting and analyzing epidemiologic information for rare diseases, which is limited and often relies on manual, error-prone methods, hindering better public health decisions.
• Researchers created a specialized corpus and developed a deep learning framework, EpiPipeline4RD, involving a Web interface and API, to automatically identify and extract epidemiologic data from scientific abstracts using a fine-tuned BioBERT model.
• Their model achieved high accuracy, demonstrated through quantitative metrics (F1 scores over 0.81) and qualitative comparisons, showcasing the potential for scalable and precise extraction of rare disease epidemiology information to support research and policy-making.