Introduction of MedModels
summary
Claims that a computer scientist will win a Nobel in medicine within 15 years and that machine learning can unlock longitudinal EHR and claims data are backed by a concrete stack: a Python package called MedModels that implements medical concept embeddings (skip‑gram style vectors learned from event sequences), treatment‑effect estimation and patient matching, sequential predictive models, and synthetic EHR generation for privacy. It targets millions‑patient datasets, is in active research and client pilots, and aims to replace repeated one‑off implementations with an engineered, auditable toolkit.
AI-written summary · about
chapters (6)
- 0:00Start
- 1:42Where do Patient Data come from?
- 3:22Rapidly increasing research interest¹
- 3:57Practical application lags behind
- 7:15Medical Concept Embeddings¹
- 9:02Applying NLP knowledge helps
related talks
28mThe Future of Healthcare – AI’s Promise and the Challenges Ahead
33mLessons Learned Applying Large Language Models in Healthcare
20mDiagnostic Medical application of ML
20mBERT Model for Real World Healthcare Data
20mAdvancing Epigenetic Predictors with Scalable Machine Learning
09mHow AI is changing the healthcare system
58mHealthcare with AI of the Future
27mXavier Amatriain, Cofounder & CTO, Curai