You have found the cui2vec explorer. This website will let you interact with embeddings for over 108,000 medical concepts. These embeddings were created using insurance claims for 60 million americans, 1.7 million full-text PubMed articles, and clinical notes from 20 million patients at Stanford. More information about the methods used to create these embeddings can be found in our preprint: https://arxiv.org/abs/1804.01486
This tab displays an interactive t-sne visualization of all 108,477 concepts.
This tab will allow you to sort all concepts by similarity to a query concepts. You can filter by semantic type, which will allow you to answer questions such as "What drug is most related to pancreatic cancer?" or "What symptoms are associated with malaria?".
Below is a concept map for each embedding created by cui2vec. The map was created by using t-sne to map the 500-dimensional vectors into 2-dimensional space. The points are colored by the data source in which they were found, and 16,299 points in blue were found in multple sources. The map is fully interactive and will give information on hover and allows you to zoom into certain regions of the plot with your mouse. Clicking on the legend will allow you to hide data from the different sources.
Note: The plot takes a few seconds to load initially, so please be patient.