NATO Hackathon Winner: AI-Powered Drug Search
summary
A NATO hackathon winner built a five-day mobile/web app that maps 700,000 drug brand names to 34,000 active ingredients using a DrugBank dump. The app uses PostgreSQL pg_trgm trigram fuzzy search plus LIKE/ILIKE ranking and adds vector search with Azure AI embeddings stored in PostgreSQL; it also supports OCR and barcode scanning to link scans to records and uses patient QR/NFC tokens for patient info. Ukrainian NGOs are already using the deployed webapp for sorting and medication matching.
AI-written summary · about
chapters (13)
- 0:00NATO Hackathon win
- 2:12Drug search task
- 5:34Medic consultations
- 9:14Mobile app demo
- 10:41Categories & storage
- 16:05Patient secure profile
- 19:00Evacuation logging
- 24:47Tech stack
- 29:27Drug search methods
- 36:02Vector search chatbot
- 38:58OCR drug scanning
- 40:01App setup & usage
- 41:29Q&A
related talks
28mAdvancing Drug Search with PostgreSQL and Azure AI
15mKnowledge Graph of Drugs Data for Swiss Healthcare System
11mAI Based Patent Search For Chemical Compound & Drug Discovery
35mCan AI solve the biggest problem in drug discovery? Time!
26mExtracting Medical Attributes from Clinical Trials
28mAI in healthcare - Deep learning for precision psychiatry
47mAI for the Proper Use of Medications
27mWhat’s in a Name? – Fuzzy Matching Techniques for Proper Nouns