Extracción automática de entidades médicas desde documentos digitales relacionados a la salud

Abstract

Abstract: Digital medical documents with a narrative format store information on patients' diseases, procedures and medications. By having an unstructured format, it is necessary to have tools that allows this information to be automatically extracted and thus avoid the manual work that experts should do. With this problem, a methodology was developed that allows the automatic extraction of medical entities from documents written in Spanish, using free access natural language processing (NLP) tools such as METAMAP and cTAKES to achieve effective management of the information contained in medical documents that automatically map, codify and normalize, according to a standard medical vocabulary, all the medical entities that can be found in the texts. In addition, an automatic process of translation and spelling correction is carried out to make the most of the potential of the tools that are developed to work mainly with documents written in English.