Knowledge Graph of Drugs Data for Swiss Healthcare System
summary
A bachelor thesis builds a Neo4j knowledge graph of Swiss and European public drug datasets using R to produce CSVs and loads them into Neo4j, staying well under the 200k node limit. Practical findings include avoiding one‑to‑one table→label lifts by keeping common attributes on nodes, replacing join tables with direct relationships to enable cluster analysis, storing arrays as node properties and using UNWIND, and resolving mismatched ATC codes by creating indirect relationship edges. Visual work uses Neo4j Bloom with attribute flags and rule based coloring, and warns that perspective versioning breaks deep links.
AI-written summary · about
chapters (13)
- 0:00Introduction and speaker background
- 1:30Why I started with graphs
- 2:30Data sources and initial model
- 3:15Getting started with Neo4j
- 4:44Mindset shift for graph data
- 5:26Data modeling pitfalls
- 6:34Relationships and cluster analysis
- 7:38Handling untidy data
- 8:44Enriching data from multiple sources
- 10:31Visual analysis challenges in Bloom
- 11:58Workarounds for Bloom visualization
- 13:02Neo4j Bloom tips and deep links
- 14:05Summary and resources
related talks
15mEnhancing Medical Data Processing With Neo4j
24mUnleashing the power of FHIR data with knowledge graphs
24mWhy You Need a Knowledge Graph And How to Build It
42mFighting fraud with graph databases
46mGraph Database
26mUsing Dbpedia and Wikidata Knowledge Graphs With Neo4j
25mKnowledge graphs for financial institutions
30mBeyond the Silos Achieving Data Harmony With Graph Based ER