Natural Language Processing for Healthcare: Current Trends...
summary
Overview of concrete claims, tools and failures: clinical NLP uses BERT-family models with many domain variants like BioBERT and ClinicalBERT, and the big leap is GPT-3 with 175 billion parameters and an estimated $12M training cost for general language tasks; practical work on MIMIC‑III shows off‑the‑shelf tokenizers fail while stanza outperforms spaCy/NLTK on clinical parsing. Pipelines rely on careful preprocessing, active/federated learning for tuning, and warn that false positives/negatives have real patient risk.
AI-written summary · about
chapters (12)
- 0:00Start
- 0:50Introduction about Nlp What Is Nlp
- 2:44Data Mining
- 3:11Language Modeling and Translation
- 5:15Technology Highlights
- 13:51Common Tasks of Application of Nlp in the Healthcare
- 14:27Classification
- 15:54What Is a Typical Nlp Pipeline for Healthcare Applications
- 17:07Sentence Parsing
- 19:56Downstream Tasks
- 20:43Computer Assisted Coding
- 23:13Federated Learning
related talks
17mNatural Language Processing in Automation
33mLessons Learned Applying Large Language Models in Healthcare
06mAdvancements in Natural Language Processing (NLP) and Text Analytics
20mBERT Model for Real World Healthcare Data
26mNatural Language Processing (NLP) going mainstream
58mHealthcare with AI of the Future
23mLeveraging Generative AI and Transformers in Healthcare
30mA library a day keeps the doctor away