Architectural Design and Development of a configurable cloud Clinical Natural Language Processing (CNLP) System

Significant amount of valuable clinical information are captured in narrative, unstructured free text in clinical reports and Electronic Medical Records (EMRs). Clinical data in free textual form is not standardized for interoperability or in a computer accessible form for manipulation. Clinical Natural Language Processing (CNLP) systems employ sophisticated linguistic and syntactical approaches to extract medical phrases contained in free text and map them to biomedical concepts in controlled medical terminologies that contain clinical terms that are representative of information contained in free text. This thesis proposes to address the barriers to clinical adoption of NLP by architecting web based configurable multi-terminology cloud Clinical NLP (CNLP) system with access to multiple CNLP engines. Input phrases can be auto-coded using either the Unified Medical Language System (UMLS) or Systematized Nomenclature of Medicine - Clinical Terms (SNOMED CT) standardized terminology systems and can be submitted to different NLP systems - MojoMapper and MetaMap. The non-technical NLP user can access this functionality remotely via a user-friendly interface. The outcome of this work will alleviate some challenges of clinical adoption of NLP.