https://github.com/dermatologist/cui-embed
Gensim wrapper for the clinical (CUI) embeddings
https://github.com/dermatologist/cui-embed
cui embeddings gensim health-informatics
Last synced: 9 days ago
JSON representation
Gensim wrapper for the clinical (CUI) embeddings
- Host: GitHub
- URL: https://github.com/dermatologist/cui-embed
- Owner: dermatologist
- Created: 2020-03-20T01:22:40.000Z (over 5 years ago)
- Default Branch: master
- Last Pushed: 2023-11-13T20:49:58.000Z (almost 2 years ago)
- Last Synced: 2025-08-14T12:41:39.611Z (about 2 months ago)
- Topics: cui, embeddings, gensim, health-informatics
- Language: Python
- Homepage: http://nuchange.ca
- Size: 15.6 KB
- Stars: 3
- Watchers: 2
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# cui-embed
## Update 2/16/2023
* Added support for LOINC embedding from [this repo](https://github.com/elleros/DSHealth2019_loinc_embeddings). Please cite [the paper](https://arxiv.org/abs/1907.09600) in the above repo if you use LOINC embedding.
This is a gensim wrapper for the Concept Unique Identifier (CUI) embeddings from the model presented [here](https://github.com/beamandrew/cui2vec). The paper describing the model creation is [here](https://arxiv.org/abs/1804.01486). This can be used with [pyomop](https://github.com/dermatologist/pyomop) for efficient electronic phenotyping from [OHDSI](https://www.ohdsi.org/) CDM databases.
You can use [cui-cdm](https://github.com/E-Health/cui-cdm) for mapping CUI to OHDSI OMOP concepts and [umlsjs](https://github.com/dermatologist/umlsjs) for accessing UMLS REST APIs for CUI and other terminology management.
* This does not include data/model.
* Model is downloaded on first use
### Install
```
pip install git+https://github.com/dermatologist/cui-embed.git
```
### Use
```
# CUI embedding
from cui_embed import Cuimodel
cm = Cuimodel('cui')
m = cm.model()
print(m.most_similar_cosmul(['C0002268']))
# LOINC embedding
lm = Cuimodel('loinc')
lm = cm.model()
print(lm.most_similar_cosmul(['2324-2']))
```