mediKanren: A System for Bio-medical Reasoning
summary
A system called mediKanren combines miniKanren relations, a GUI, and a custom on‑disk graph store to run low‑latency biomedical queries; data are stored as Racket s‑expressions in giant vector files with RAM indexes so the engine can jump to any record. Engineering yields a compact, modifiable miniKanren implementation hooked to greg db and it handles billions of assertions. A deployed query pipeline surfaces plausible links (example: imatinib treats asthma) and helped suggest treatment leads for a rare genetic case.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 2:32Why mediKanren
- 3:40Biolink Model
- 9:05Representation
- 10:30Graph Database
- 13:12Data Representation
- 14:30Query
- 16:00Flexibility
related talks
44mA Vision for Relational Programming in miniKanren
23mImproving stableKanren’s Backward Compatibility
43mPost Functional
16mMedQGraph: TMKGs for AI-Driven Healthcare Insights
22mThe Nexus of Innovation: How Knowledge Graphs Fuel AI Scientific Breakthroughs
25mArtificial intelligence with Elixir the Rete (...)
13mBuilding High-Performance AI Literature Processing
53mThe Algorithm for Precision Medicine