ABSTRACT
Health-related data are often stored in the form of descriptive documents that are highly unstructured, making it hard for machines to understand. Clinical Natural Language Processing (CNLP) systems retrieve relevant data from unstructured medical narratives, interpret their linguistic features, identify the meaning of biomedical terms, and convert these terms into a form best understood by the machines for clinical decision-making. This research reviews the current CNLP systems, focussing on the various NLP phases and their clinical applications. The existing CNLP systems that capture structured data from the unstructured free text were identified using the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) approach. An exclusive survey on numerous NLP techniques for healthcare was carried out to select the popular CNLP systems for extracting clinical terminology from medical narratives. Four research databases were searched using a query integrating NLP and structured data analysis from 2006 to 2021. The literature was narrowed down for each information retrieval system, and the Mendeley software's “Check for Duplicates” function checked all fields for possible duplicates and eliminated them. A manual scrutinization of the remaining article's titles and abstracts was performed to acquire the final set of articles. This article gives the basics of NLP, its phases, and applications in the clinical domain and finally reviews the list of popular CNLP systems available for extraction of key medical terms from clinical notes and can be a value addition for the acceleration of research in CNLP systems.
