Word embedding
A fixed-length vector of trainable numbers that represents a word so a computer can work with it. Related words end up with similar values in some dimensions: apple sits near orange, both share something with juice, while airplane sits far from all of them.
A hospital upgrades its intranet search so staff can find policies by meaning instead of exact wording. Words that appear in similar sentences end up with similar embeddings, and opposites such as hypoglycemia and hyperglycemia, or hypokalemia and hyperkalemia, appear in very similar sentences. A search for the hypoglycemia protocol can then rank the hyperglycemia protocol near the top. Before rollout, test the search with pairs of opposite clinical terms and check that the right document comes first.
Terms like this come up in real clinical scenarios across the HelloAI courses: bite-sized modules with verifiable certificates. An account takes one minute, no password needed.
Sign in →