NEJM AI Grand Rounds Translational AI in Medicine: Unlocking AI’s Potential in Health Care with Nigam Shah
38 snips
May 15, 2024 Dr. Nigam Shah, a distinguished Professor of Medicine at Stanford University, discusses the transformative impact of computational tools in healthcare, the integration of AI into clinical settings, the importance of data quality in training AI systems, and the democratization of medical knowledge. The conversation also explores the role of open-source models, the application of AI in administrative tasks at Stanford healthcare, and the potential of large language models in medicine.
AI Snips
Chapters
Transcript
Episode notes
AI Implementation Strategy
- AI implementation in healthcare often prioritizes low-risk administrative tasks over direct patient care initially.
- This approach allows for safer experimentation and faster deployment, but research should explore complex diagnostic problems.
Productivity vs. Efficiency
- Healthcare prioritizes productivity (more care delivered per resource) over efficiency (reducing workload without increasing output).
- AI solutions focusing solely on efficiency may not improve overall healthcare productivity.
LLM Basics
- Large language models (LLMs) predict the probability of word sequences based on training data, generating text or embeddings (numerical representations).
- LLMs can process various sequential data types beyond human language.