Natural Language Processing for Automated Extraction of Medical Terms in Electronic Health Records

Abstract

This research aims to explore application of NLP approaches for the automatic extraction of medical
terminologies from Electronic Health Records (EHRs). A system to automate clinical relevant information
retrieval through named entity recognition (NER) and transformer models was created. The solution was
applied to classical biomedical data and was tested against multiple other systems. The findings showed
significant differences in the precision and recall results that were computed. This claim of concept
validates the degree of improvements that can be achieved by using NLP on EHRs which are essential
parts of clinical decision support systems and automatic data entry systems.