Natural Language Processing (NLP)

Clinical documentation used in coding and charge capture comes from a variety of sources and in variety of formats. In most cases, clinical documentation is a disparate collection of contents ranging from free text, unstructured text, HL7 messages and template driven structured content. Such disparate contents are derived from a distributed array of systems owned by different departments including pharmacy, radiology, lab, respiratory, ER and physician groups with little similarity with each other.

Medical Savant incorporates a natural language processing engine to process such disparate collection of clinical documentation. Conventional NLP engines enable lexical parsing and extraction of grammatical concepts such as verb and noun, and general concepts such as names, dates and places. Unlike conventional NLP engines that are optimized for lexical parsing, Medical Savant NLP engine is a clinically cognizant concept (C3) extraction platform. The NLP enables:

  • Assembly of content from a variety of sources and formats; parse,
  • Demarcation and contextual grouping sections of document – a.k.a. tagging;
  • Extraction of clinically relevant phrases and concepts from the document;
  • Semantic categorization of clinical phrases and concepts such that the various coding engines can execute the appropriate coding algorithms.