Discovery of Biomedical Databases Through Semantic Questioning.

Area of Science:

• Biomedical Informatics
• Data Science
• Health Data Management

Background:

• Clinical studies require efficient identification of relevant datasets from health data catalogues.
• Current metadata harmonization issues create bottlenecks in searching distinct biomedical databases.
• Question-answering interfaces can optimize data exploration but are limited by data heterogeneity.

Purpose of the Study:

• To present a methodology for semantic search across multiple biomedical database catalogues.
• To enable efficient extraction and utilization of information using shared domain knowledge.
• To facilitate the publication of converted data as FAIR (Findable, Accessible, Interoperable, Reusable) endpoints.

Main Methods:

• Developing a pipeline for semantic data extraction from various biomedical catalogues.
• Utilizing a shared domain knowledge base to harmonize disparate metadata.
• Implementing an end-user interface that accepts natural language queries for data retrieval.

Main Results:

• Successful semantic search capability across multiple, previously unharmonized, biomedical databases.
• Converted data made available as FAIR endpoints, enhancing data accessibility and reusability.
• An intuitive natural language interface for researchers to query complex health datasets.

Conclusions:

• The proposed methodology overcomes metadata harmonization challenges in biomedical data discovery.
• Enables more efficient and effective identification of relevant datasets for clinical research.
• Promotes data sharing and utilization through FAIR data principles and natural language querying.