Hyper-v

Lecture19 – EHRs – MLCB24



Playlist:
Notes:
Lecture Slides:
Lecture Chapter: TBD

00:00 Intro and Overview of the Lecture
08:59 Dealing with Population-Scale EHRs
16:29 Dealing with Multiple Phenotypes
21:40 Mantis AI for Browsing Patient Latent Spaces
25:55 MixEHR for Latent Dirichlet Allocation modeling of Patient Subtypes
29:05 Missing Mechanism for Data Not Missing at Random (NMAR)
37:04 PheWAS across Multiple Phenotypes and Reversing GWAS
40:25 Causal Inference with Genetic Conditioning for BMI-TG-T2D-LDL-CAD-Height
41:52 EHR Interventions, Biomarkers, Risk, Treatments, Reverse Causality
48:07 Beyond Structured Data – Natural Language Processing NLP foundations
48:35 Classical NLP – Extracting structured information from clinical notes
53:35 Grammar, Syntax, and Simplified Linguistics
54:55 Term spotting + handling negation, uncertainty
56:10 Expanding terms with Synonyms, Semantic Networks, Term Relationships
58:39 pre-NN ML to identify entities and relations
59:13 Topic modeling with LDA
1:00:52 Transition to Language Models (LLMs)
1:03:21 Context, Co-occurrence, Vector space embeddings, and Word2Vec
1:08:40 Tokenization, Words, Sub-words, Characters
1:09:09 Context, LSTMs, Memory, Phrases, and Transformers
1:10:11 Self-Attention with Transformers, GTP-2/3
1:13:19 GPT-4 – Actions, Agents, Reinforcement Learning
1:16:13 AI Agents in Value-Based Healthcare
1:27:43 Summary and Outro

[ad_2]

source

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button