Some friends and I wanted to experiment with whether you could do natural language processing in the client. We built a clinical NLP engine that takes text and identifies disease mentions with their associated ICD-10-CM codes. It is fast (efficient hash + no need to send off to a server) and safe since no health information is ever sent over the internet. We were happy with the result. Let us know what you think!
We believe the tool could be useful for clinical research (rapid cohort identification), medical auditing, and more.