Knowledge Combined Common Data Model: A Graph-Based Integration of Reactome and OMOP CDM
summary
Unified graph loads Reactome and an OMOP CDM-converted MIMIC‑IV into Neo4j and adds ontology links from BioOntology so queries traverse from patient records to molecular pathways. The graph contains 33 million nodes, 64 million relations, including 337,000 patients, 17 million drug exposures, 4.66 million condition occurrences. Mapping finds 184 Reactome chemical→CDM drug links and 574 DO→CDM disease links. Cypher queries list patients prescribed a molecular entity or follow a drug node into pathway reactions to show putative mechanisms.
AI-written summary · about
chapters (11)
- 0:00Introduction and speaker intro
- 1:55Speaker background and KCDM overview
- 4:54Problem with clinical and biological data silos
- 7:26Building the KCDM graph structure
- 10:27Graph size and data composition
- 12:19Graph schema and data sources
- 14:36Querying the integrated graph
- 19:21Example query from Reactome to CDM
- 22:14Future expansions of KCDM
- 23:43Summary and precision medicine potential
- 25:15Q&A and availability
related talks
15mEnhancing Medical Data Processing With Neo4j
36mFormatting Biological Big Data to Enable (Personalized) Systems Pharmacology
33mUnderstanding Molecular Complexity for Precision Medicine
30mContext Engineering: Connecting the Dots with Graphs
42mGraph Theory Takes on the Opioid Epidemic
24mUnleashing the power of FHIR data with knowledge graphs
33mNetwork Data Integration Enables Precision Medicine
14mOntology-Backed GraphRAG: Injecting Biomedical Logic in LLMs for Drug Discovery