MedKnowts: Unified Documentation and Information Retrievalfor Electronic Health Records
summary
A deployed prototype unifies note editing and search for Emergency Department scribes, capturing structured data via autocomplete chips (rule-based triggers plus a 1‑D neural predictor and slash filters) and post‑recognition using a modified Aho‑Corasick; it auto‑populates later sections with a greedy algorithm and modified negex. Year‑long use covers 1,185 patients and a month‑long deployment with four scribes covers 234 patients, during which the system proactively displays 3,614 cards and users report faster documentation with role‑dependent adoption and manual disambiguation of ambiguous terms.
AI-written summary · about
related talks
39mContent Management as Critical Success Factor for Personalized Medicine
20mKnowledge Scout: From Talks and Slides to Smart Summaries
34mData Query, US Core
16mCognitive A11y: Designing Neuroaccessible docs for Neurodivergent minds
31mDoc Health Metrics: How Two Writers Built a Framework to Prioritize Docs
30mThe Blue Side of Documentation
16mInteract with Your Technical Documentation via Generative AI
10mCrowdsourcing Scholarly Discourse Annotations