Querying semantic catalogues of biomedical databases.

Area of Science:

• Biomedical Informatics
• Health Data Science
• Observational Studies

Background:

• Secondary use of health data is crucial for medical and biomedical discoveries.
• Multi-center observational studies face challenges in identifying suitable datasets.
• A deep understanding of individual datasets is often required, hindering research progress.

Purpose of the Study:

• To develop a strategy for retrieving semantically annotated biomedical datasets.
• To enable complex data queries using natural language interfaces.
• To overcome the limitations of traditional data retrieval methods in multi-center studies.

Main Methods:

• Transforming natural language questions into formal language queries.
• Semantically annotating biomedical datasets using open biomedical ontologies.
• Developing a natural language interface for querying semantic data.

Main Results:

• Validated methodology using Alzheimer's disease datasets on a European data-sharing platform.
• Published data as a FAIR endpoint in a semantic information format.
• Demonstrated the ability to handle single-concept, exclusion criteria, and multi-concept questions.

Conclusions:

• The proposed strategy facilitates the use of biomedical data through semantic transformation and natural language querying.
• This approach enhances the accessibility and utility of health data for observational research.
• The method simplifies complex data retrieval without requiring users to master formal query languages.