Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Natural Language Processing”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 73 records · Page 4Linked to original sources
PubMed1999

Restricted natural language processing for case simulation tools.

For Interactive Patient II, a multimedia case simulation designed to improve history-taking skills, we created a new natural language interface called GRASP (General Recognition and Analysis of Sentences and Phrases) that allows students to interact with the program at a higher level of realism. Requirements included the ability to handle ambiguous word senses and to match user questions/queries to unique Canonical Phrases, which are used to identify case findings in our knowledge database. In a simulation of fifty user queries, some of which contained ambiguous words, this tool was 96% accurate in identifying concepts.

Computer Simulation↗
PubMed2002

A study of the medical record interface to natural language processing.

The information about a patient tends to be handled more on a computer system. However, it is not sufficiently rational enough because of the fundamental difference between man and a computer. Up to now, man has treated information using a natural language. Therefore, if it can be applied to handle medical information electrically, that will become more rational. For this reason, we developed a new classification method that interfaces a computer with the human being, using a natural language. This method was named as a situation-oriented medical record, and this depicts changes in a situation by the case frame from a viewpoint of man's cognition. Moreover, the medical communication by a natural language, which is currently used when a patient changes a hospital, was analyzed in order to confirm the validity of this method. In addition, we developed a prototype system that allows computers to implement this kind of communication.

Humans↗
PubMed2000

Extending the VA CPRS electronic patient record order entry system using natural language processing techniques.

An automated practitioner order entry system was recently implemented at the VA Puget Sound Health Care System. Since the introduction of this system, we have experienced various problems, among them an increase in time required for practitioners to enter orders. In order to improve usability and acceptance of the order entry, an alternate pathway was built within CPRS that allows direct natural language based order entry. Implementation of the extension in CPRS has been made possible because of the three layers CPRS architecture and its strong object oriented models. This paper discusses the advantages and needs for a natural language based order entry system and its implementation within an existing order entry system.

Humans↗
PubMed2002

Using natural language processing to analyze physician modifications to data entry templates.

Efficient data entry by clinicians remains a significant challenge for electronic medical records. Current approaches have largely focused on either structured data entry, which can be limiting in expressive power, or free-text entry, which restricts the use of the data for automated decision support. Text-based templates are a semi-structured data entry method that has been used to assist physicians in manually entering clinical notes, by allowing them to edit predefined example notes. We analyzed changes made to 18,726 sentences from text templates, using a natural language processor. The most common changes were addition or deletion of normal observations, or changes in certainty. We identified common modifications that could be captured in structured form by a graphical user interface.

Medical Records Systems, Computerized↗
PubMed2005

Empirical data on corpus design and usage in biomedical natural language processing.

This paper describes the design of six publicly available biomedical corpora. We then present usage data for the six corpora. We show that corpora that are carefully annotated with respect to structural and linguistic characteristics and that are distributed in standard formats are more widely used than corpora that are not. These findings have implications for the design of the next generation of biomedical corpora.

Computational Biology↗
PubMed2001

HPARSER: extracting formal patient data from free text history and physical reports using natural language processing software.

A prototype, HPARSER, processes a patient history and physical report such that specific data are obtained and stored in a patient data record. HPARSER is a recursive transition network (RTN) parser, and includes English and medical grammar rules, lexicon, and database constraints. Medical grammar rules augment the grammar rule base and specify common phrases seen in patient reports (e.g., "pupils are equal and reactive"). Each database constraint associates a grammar rule with a database table and attribute. Constraint behavior is such that if a rule is satisfied, data is extracted from the parse tree and stored into the database. Control reports guided construction of grammar and constraint rules. Test reports were processed with the control rules. 85% of test report sentences parsed and a 60% data capture rate, compared to controls, was achieved. HPARSER demonstrates use of an RTN to parse patient reports, and database constraints to transfer formal data from parse trees into a database.

Humans↗
PubMed2001

Comparing general and medical texts for information retrieval based on natural language processing: an inquiry into lexical disambiguation.

In this paper we compare two types of corpus, focusing on the lexical ambiguity of each of them. The first corpus consists mainly of general newspaper articles and literature excerpts, while the second belongs to the medical domain. To conduct the study, we have used two different disambiguation tools. First, each tool was validated in its respective application area. We then use these systems in order to assess and compare both the general ambiguity rate and the particularities of each domain. Quantitative results show that medical documents are lexically less ambiguous than unrestricted documents. Our conclusions emphasize the importance of the application area in the design of NLP tools.

Information Storage and Retrieval↗
PubMed2006

Extracting principal diagnosis, co-morbidity and smoking status for asthma research: evaluation of a natural language processing system.

BACKGROUND: The text descriptions in electronic medical records are a rich source of information. We have developed a Health Information Text Extraction (HITEx) tool and used it to extract key findings for a research study on airways disease. METHODS: The principal diagnosis, co-morbidity and smoking status extracted by HITEx from a set of 150 discharge summaries were compared to an expert-generated gold standard. RESULTS: The accuracy of HITEx was 82% for principal diagnosis, 87% for co-morbidity, and 90% for smoking status extraction, when cases labeled "Insufficient Data" by the gold standard were excluded. CONCLUSION: We consider the results promising, given the complexity of the discharge summaries and the extraction tasks.

Asthma↗
PubMedDate not supplied

Natural language processing and the representation of clinical data.

OBJECTIVE: Develop a representation of clinical observations and actions and a method of processing free-text patient documents to facilitate applications such as quality assurance. DESIGN: The Linguistic String Project (LSP) system of New York University utilizes syntactic analysis, augmented by a sublanguage grammar and an information structure that are specific to the clinical narrative, to map free-text documents into a database for querying. MEASUREMENTS: Information precision (I-P) and information recall (I-R) were measured for queries for the presence of 13 asthma-health-care quality assurance criteria in a database generated from 59 discharge letters. RESULTS: I-P, using counts of major errors only, was 95.7% for the 28-letter training set and 98.6% for the 31-letter test set. I-R, using counts of major omissions only, was 93.9% for the training set and 92.5% for the test set.

Diagnosis, Computer-Assisted↗
PubMed2003

Building and evaluation of a structured representation of pharmacokinetics information presented in SPCs: from existing conceptual views of pharmacokinetics associated with natural language processing to object-oriented design.

OBJECTIVE: Develop a detailed representation of pharmacokinetics (PK), derived from the information in Summaries of Product Characteristics (SPCs), for use in computerized systems to help practitioners in pharmaco-therapeutic reasoning. METHODS: Available knowledge about PK was studied to identify main PK concepts and organize them in a preliminary generic model. The information from 1950 PK SPC-texts in the French language was studied using a morpho-syntactic analyzer. It produced a list of candidate terms (CTs) from which those describing main PK concepts were selected. The contexts in which they occurred were explored to discover co-occurring CTs. The regrouping according to CT semantic types led to a detailed object-oriented model of PK. The model was evaluated. A random sample of 100 PK texts structured according to the model was judged for completeness and semantic accuracy by 8 experts who were blinded to other experts' responses. RESULTS: The PK text file contained about 300000 words, and the morpho-syntactic analysis extracted 17520 different CTs. The context of 592 CTs was studied and used to deduce the PK model. It consists of four entities: the information about the real PK process, the experimental protocol, the mathematical modeling, and the influence of factors causing variation. Experts judged that the PK model represented the information in 100 sample PK texts completely in 89% of cases and nearly completely in the other 11%. There was no distortion of meaning in 98% of cases and little distortion in the remainder. CONCLUSION: The PK model seems to be applicable to all SPCs and can be used to retranscribe legal information from PK sections of SPCs into structured databases.

Artificial Intelligence↗