You are 
Home 
News Science and research TeSSI® Natural Language Processing for Healthcare and Life Sciences Knowledge Discovery
Home News Science and research TeSSI® Natural Language Processing for Healthcare and Life Sciences Knowledge Discovery 
News feeds
Email news|
TeSSI® Natural Language Processing for Healthcare and Life Sciences Knowledge Discovery |
| Written by Kyle Silvestro | |
| Wednesday, 28 March 2007 | |
|
Natural Language Processing powered by the world's largest Medical Ontology "NLP is only as good as the Knowledge base (Ontology) that supports it"
Semantic Indexing, Searching and Retrieval (Queries in Natural Language) Information Extraction to feed Clinical Information Systems Document Categorization and Clustering engine using Semantic Technology Processing Unstructured Information Clinicians and researchers struggle with an ever increasing information overload in the form of journal articles, practice guidelines, patient notes, discharge summaries, etc. The information in these sources is generally presented in the form of unstructured text. Many vendors try to solve this problem by offering solutions where physicians enter structured information using pick lists based on Controlled Medical Vocabularies (CMV). Physicians, however, do not like to be restricted in terminology but prefer to use free text to record their findings in natural language. Also, medical research and reports will more than likely always be documented in a free text format. CMVs are not designed to assist in the automated processing of medical free text. Other solutions to automated processing of unstructured clinical information are based on medical classification systems. These systems however are not capable of dealing with unstructured information effectively, because they simply are not designed to ‘understand’ medical language. Indexing, searching and retrieving of information will be substandard, and automatically extracting information will not be possible at all. TeSSI®: semantic processing of free text information L&C's TeSSI® middleware suite leverages state-of-theart Natural Language Processing (NLP) technology to solve the problems associated with automated processing of unstructured clinical information. The TeSSI® components deliver a higher degree of accuracy for indexing, search and retrieval, and categorization applications. Additionally, TeSSI® delivers the world's first automated information extraction solution for the medical domain. To achieve this high degree of accuracy, TeSSI® uses the intelligence stored in L&C's LinKBase®, a conceptual computer-understandable representation of medicine and the only medical knowledge base specifically developed for purposes of NLP in the medical domain. LinKBase® is a formal medical ontology containing over 1,500,000 concepts, linked together into a network using more than 480 distinct relationship types. Each concept is also linked to a number of different terms in English and other languages. The TeSSI® middleware suite consists of four components that can be installed separately or can be combined to offer an innovative, complete semantic information management environment: TeSSI® Indexing Engine, TeSSI® Search Engine (with optional the Clustering and Categorization engine) and TeSSI® Extraction Engine. TeSSI® Indexing Engine When other companies talk about 'semantic indexing' of documents, they really don't mean anything more than indexing based on keywords or statistics. Semantic knowledge is definitely not applied. TeSSI® creates indexes based on semantic information, i.e. the underlying meaning of words and phrases. TeSSI® identifies medical terms within the text, looks up the medical concepts behind these terms in the LinKBase® ontology, and places the concepts (or if desired codes such as SNOMED or ICD-9 and ICD-10 that are linked to the concepts) into the index. TeSSI® then uses contextual information to determine how relevant each concept is to the document as a whole. During the indexing process, TeSSI® identifies contextual information such as word variations, negations (see the example below), modality and disambiguation. TeSSI® Search Engine Semantic Search powered by TeSSI® allows a knowledge seeker to query a clinical content store in his own native language and still retrieve an accurate list of highly relevant documents. While the TeSSI® Indexing Engine creates semantic indexes to add to unstructured data, the TeSSI® Query Analyzer performs a transformation of a user query in natural language into its conceptual representation using the intelligence in the LinKBase® ontology. Once this transformation has been performed, the resulting conceptual index is matched to the conceptual document indexes to identify and retrieve the relevant documents with great accuracy. L&C's advanced NLP technology greatly increases recall and precision of search and retrieval functions for both external portals and corporate intranets. Retrieved documents can then further be categorized by using the optional Clustering and Categorization Engine.
TeSSI® Extraction Engine The TeSSI® Extraction Engine is the only tool available that accurately extracts clinical facts from unstructured clinical documents (discharge summaries, pathology reports, etc.). Extracting information from full text in natural language is becoming increasingly important in the healthcare delivery and the pharmaceutical industries. Rapid access to information in a pre-configured format is the "holy grail." In contrast, manual access to particular information is difficult for several reasons: o Documents such as medical discharge summaries contain superfluous information that is not relevant for particular purposes; o Medical documents are rather long and time consuming to read; o Differences in style among physicians means that where information is placed and how it is expressed can vary from document to document. o The number of documents from which information has to be extracted makes it impossible to do this manually. The purpose of L&C’s Information Extraction (IE) technology is to automatically extract user-defined information to a user-defined format, using semantic information. Information Extraction is the process of filling-in the blanks of pre-configured templates with relevant information from human-readable text. IE thus involves identifying the text fragments that contain the desired information and insertion of this information into the correct template blank. The input is an empty template and a set of texts, and the output is a set of completed templates in XML, a database compliant format. With the TeSSI® Extraction Engine, companies now have a powerful tool to make the vast amount of important clinical facts -- previously buried in unprocessable free-text documents -- available for further processing. In pharmaceutical research, for example, rapid access to information speeds up the drug discovery and development process in all its phases and decreases the time to market of new drugs plus reduces the risk of failure. TeSSI® Categorization Engine Semantic indexes provide the information needed for document categorization. Categories may be predefined or dynamically generated based on ranking the importance of document concepts. Combined with document retrieval capabilities, for instance, categorization allows more intuitive display of search results and reduces navigation time. About Language and Computing Language and Computing Inc. (L&C) has invested over 10 years in the research and development of the LinKSuite® solution. The company was incorporated in 1998 to market medical natural language processing solutions designed to transform information management in the healthcare industry. Today, Language and Computing has a highly qualified staff of PhDs and MDs with degrees in computer science and linguistics, and experienced software engineering and sales/ marketing teams to service and support customers. Language and Computing has offices in both North America and Europe. LANGUAGE AND COMPUTING Inc. North American Headquarters 11654 Plaza America Drive, #716 Reston, VA 20190 USA (888) 579-0682 (Toll Free) |
| < Prev | Next > |
|---|
Sponsored Links
.
Agriculture Automotive Books / Publishing Books /Publishing Business Computer Consumer Design Education Employment/Careers Engineering Environment Events/ Trade shows Finance/Money Gaming Government Home and Family Industry Insurance Latest News Legal and Law Lifestyle Maritime Media Medical Misc Music News Non Profit Politics Religion Science and research Self Help Society Sports Technology Telecoms News Trade Transportation Travel/Hospitality Wine
Created with AkoCloud 1.1 final.
Bookmark with:
What are these?