DrNote: An open medical annotation service.

Area of Science:

• Medical Informatics
• Natural Language Processing
• Bioinformatics

Background:

• Medical text mining offers valuable insights from unstructured data in clinical research.
• Existing tools primarily support English, with limited options for non-English medical text resources.
• Practical, flexible, and easy-to-set-up tools are needed for processing diverse medical texts.

Purpose of the Study:

• To introduce DrNote, an open-source text annotation service for medical text processing.
• To provide a fast, effective, and user-friendly annotation pipeline.
• To enable custom annotation scopes and support multilingual training.

Main Methods:

• DrNote is an open-source service built upon OpenTapioca.
• It integrates publicly available datasets from WikiData and Wikipedia for entity linking.
• The system allows users to define custom annotation scopes by filtering relevant entities.
• It can be trained on any language-specific Wikipedia dataset for multilingual support.

Main Results:

• DrNote offers a complete annotation pipeline with a focus on ease of use and speed.
• The service allows for flexible, custom entity filtering for knowledge base construction.
• It demonstrates multilingual adaptability by leveraging language-specific Wikipedia datasets.
• A public demo instance is available at https://drnote.misit-augsburg.de/.

Conclusions:

• DrNote provides a practical solution for medical text mining, particularly for non-English resources.
• Its flexible and multilingual design enhances its applicability in diverse clinical research scenarios.
• The open-source nature and ease of use facilitate broader adoption in medical research.