medcodeassist

Assistant for medical coding based on analysis of free text medical documentation

API

Tokenizations

Expects

• text

Returns

• list of {word, token, pos}

Synonyms

Expects

• word
• count (of expected synonyms. Only makes sense because we are approximating synonyms by similar tokens. Will be removed when we switch to a real synonym database)

Returns

• list of simlar tokens ({name, similarity})

Note: At the moment we're approximating synonyms by using a similarity measure.

Code Proposals

Expects

• list of input_codes and input_code_types (i.e. input_codes[i] has the type input_code_types[i])
• get_drgs, get_chops, get_icds (E.g. set get_drgs to true if you want a list of drgs. You can retrieve multiple types of codes at the same time.)
• count (How many codes per type you want to retrieve)

Returns

• list of {code, similarity}

MongoDB

Run the database as follows: mongod --dbpath [...]/medcodeassist/db --rest

• This will start the mongodb server at localhost:27017
• --rest will provide an http interface at localhost:28017
• --dbpath [...]/medcodeassist/db defines your data directory

iteration1

• To see the mocked layout open /TestSite/index.html
• The front_end.html.erb is for experimental purpose
• under /spec you find tests