Healthcare "big" Data from Data Management to Web Apps
summary
Claim: healthcare "big" data is really thousands of little data problems rather than huge records. Company builds a cloud-managed on-premise appliance and a MyView EMR overlay that groups disparate sources (Hadoop back ends, a Blueprints graph DB, RabbitMQ, shared schema, core.async, Clojure/ClojureScript front and back) and deploys runtime data-as-code rules. Surprising stats: ~400 data systems in a 600–1,000 bed hospital and ~2,500 databases across systems; porting Java/Groovy to Clojure yields 4–5x fewer lines of code and faster iteration.
AI-written summary · about
chapters (21)
- 0:00Intro
- 0:58about me
- 2:21the problem we're solving
- 2:55the problem with data in healthcare
- 4:46a simple example
- 5:32how did the industry approach this?
- 7:05what is interoperability (in healthcare)?
- 8:11interpretation is highly contextual
- 10:03how do you get everyone to agree on semantics?
- 11:25our approach
- 14:00dynamic development + repl
- 16:25code as data, data as code
- 17:33JVM integration
- 19:31how we decided which libraries to use
- 21:31midje vs. clojure test
- 23:27enlive vs. hiccup/clostache came down to perspective / philosophy / style
- 24:43the journey to clojurescript
- 25:50why we want to like clojurescript
- 26:20our difficulties with clojurescript
- 27:51agile development teams
- 28:43increased code density
related talks
31mAI in healthcare: Data driven transformation
09mCan Big Data Save Them?
08mBig data answering healthcare's biggest inquiries
54mUntangling Healthcare with Spark and Dataflow
05mHadoop and Healthcare
17mHealth Care Analytics: The Key Is Integration
47mWhat it takes to process healthcare data?
12mDesigning Your Data-Centric Organization