A database of Health Condition Evolution - HES
KA pipeline
data -> RDF data in TTL format
queries -> SPARQL Anything queries for building RDF data
rawData -> database of Health Evolution Statements in CSV format
-
Corpus preparation: |_ 01trainingdataset The training dataset and preparation process
-
Knowledge components extraction: |_ 02MLAlgorithms/ Jupyter notebooks implementing (training and test process)the different algorithms used to classify sentences according to the HES. It also contains the models used to infer the HES
|_ 03predictions/ Jupyter notebooks with the application of the machine learning approach and predictions on the entire corpus
-
Knowledge completion: Jupyter notebooks with the implementation of the propagation rules For the extraction of SNOMED CT concepts we work with Snowstorm. Snowstorm is a SNOMED CT terminology server built on top of Elasticsearch, with a focus on performance and enterprise scalability. We use Snowstrom and ECL to query SNOMED CT taxonomy and obtain the list of features and relationships of each concept.
-
Human-in-the-loop Tool repository