Connecting the dots in radiation oncology data: RadPlanBio use case

Radiooncology and radiobiology research generate highly heterogeneous datasets often stored in siloed formats like spreadsheets or local databases with inconsistent terminology. This fragmentation limits data interoperability, reuse, and cross-project integration.

• Standardization: Harmonizing heterogeneous preclinical and clinical datasets using controlled vocabularies.

• Semantic Modeling: Utilizing the Ontology for Preclinical Trials in Radiation Oncology (PTRO) to map experimental entities such as demographics, treatments, and clinical assays.

• Knowledge Graph Integration: Connecting datasets across institutions (e.g., KiTZ, PCTU, MDC) to represent complex biological relationships.

• LLM Interface: Leveraging Large Language Models to translate natural-language questions into SPARQL queries, allowing researchers to interact with Knowledge Graphs without technical expertise in semantic languages.

The infrastructure focuses on refining metadata and validating LLM-supported query generation across biomedical domains to enhance access to FAIR data.