Clinics on Rails (Brighton Ruby 2026)

https://alexcwatt.com/act

https://www.youtube.com/watch?v=-PpU3xc7JJw

Transcript

1. 🧑⚕ Clinics on Rails Alex Watt
 @alexcwatt
7. 🎓 2,200 families self-supporting
9. 2025 Pittsburgh Team
11. Benefits of clinical data • Set volunteer expectations • Plan for pharmacy • Understand community health over time
12. Requirements doc January 2024
14. Surveyed options Throughout 2024
15. BackpackEMR ✅ Designed for mobile medical clinics
 ❌ PII stored outside of Ethiopia, could not host in country, limited reporting
16. Google Sheet ✅ Free, ubiquitous, simple
 ❌ Relational data, limited data validation, more difficult integrations, permissions
17. OpenMRS & others ✅ Powerful open source package
 ❌ Seemed non-trivial to deploy and customize
18. It had to be simple.
21. Built in a week (January 2025) Monday rails new Tuesday Finished data models Wednesday Deployed with Hatchbox Thursday User policies & launch Friday Addressed feedback
23. Project Impact • Volunteers can help with medical data • Medical teams see their impact • Future teams are better prepared
24. 1,583 patient encounters documented in 5 communities in Addis
25. Innovation Tokens acw.sh/boring
26. I spent only one token — on search.
27. Neoplasms Diseases of the blood or blood-forming organs Endocrine, nutritional, or metabolic diseases etc.
29. Simple Hybrid Search + semantic fuzzy match
30. Simple Hybrid Search + semantic fuzzy match Query: fizzy sparkling water soda frizzy hair Fizzy’s Diner
31. Simple Hybrid Search + informers gem (sentence-transformers) Jaro-Winkler
32. sentence-transformers/all-MiniLM-L6-v2 91 MB 1 vCPU and 1 GB RAM
33. Embedding computation MODEL_NAME = "sentence-transformers/all-MiniLM-L6-v2" EMBEDDER = Informers.pipeline("embedding", MODEL_NAME) def embedding(text) embeddings_array = EMBEDDER.(text) Vector[*embeddings_array] end
34. Embedding computation MODEL_NAME = "sentence-transformers/all-MiniLM-L6-v2" EMBEDDER = Informers.pipeline("embedding", MODEL_NAME) def embedding(text) embeddings_array = EMBEDDER.(text) Vector[*embeddings_array] end
35. Embeddings for diagnoses diagnoses_with_embeddings = Diagnosis.all.map do |diagnosis| text = "#{diagnosis.name} (e.g., #{diagnosis.examples})" embedding = embedding(text) { diagnosis: diagnosis, embedding: embedding } end
36. Query embedding QUERY_PREFIX = "Represent this sentence for searching relevant passages: " query_embedding = embedding(QUERY_PREFIX + query)
37. Search with embeddings and fuzzy text search LIMIT = 3 diagnoses_with_embeddings.filter_map do |d| semantic = cosine_similarity(query_embedding, d[:embedding]) fuzzy = fuzzy_score(query, d) semantic_relevant = semantic >= MIN_SEMANTIC_SCORE fuzzy_relevant = fuzzy >= MIN_FUZZY_SCORE next unless semantic_relevant || fuzzy_relevant [d[:diagnosis], [semantic, fuzzy].max] end.sort_by { |_, score| -score }.first(LIMIT).map(&:first)
38. Search with embeddings and fuzzy text search LIMIT = 3 diagnoses_with_embeddings.filter_map do |d| semantic = cosine_similarity(query_embedding, d[:embedding]) fuzzy = fuzzy_score(query, d) semantic_relevant = semantic >= MIN_SEMANTIC_SCORE fuzzy_relevant = fuzzy >= MIN_FUZZY_SCORE next unless semantic_relevant || fuzzy_relevant [d[:diagnosis], [semantic, fuzzy].max] end.sort_by { |_, score| -score }.first(LIMIT).map(&:first)
39. Search with embeddings and fuzzy text search LIMIT = 3 diagnoses_with_embeddings.filter_map do |d| semantic = cosine_similarity(query_embedding, d[:embedding]) fuzzy = fuzzy_score(query, d) semantic_relevant = semantic >= MIN_SEMANTIC_SCORE fuzzy_relevant = fuzzy >= MIN_FUZZY_SCORE next unless semantic_relevant || fuzzy_relevant [d[:diagnosis], [semantic, fuzzy].max] end.sort_by { |_, score| -score }.first(LIMIT).map(&:first)
42. What I hope you take away
43. #1: Your coffee ritual is (probably) weak ☕
44. #2: End-to-end ASAP 📅
45. #3: Simple software is great 👨🔧
46. #4: Open-weight models are awesome ✨
47. #5: You can just do things 💪
48. አመሰግናለሁ acw.sh/act