SEARCH · PubMed Health

Results for “Natural Language Processing”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

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

At least 1,099 records · Page 61Linked to original sources

PubMed2004

Generation of a large gene/protein lexicon by morphological pattern analysis.

The identification of gene/protein names in natural language text is an important problem in named entity recognition. In previous work we have processed MEDLINE documents to obtain a collection of over two million names of which we estimate that perhaps two thirds are valid gene/protein names. Our problem has been how to purify this set to obtain a high quality subset of gene/protein names. Here we describe an approach which is based on the generation of certain classes of names that are characterized by common morphological features. Within each class inductive logic programming (ILP) is applied to learn the characteristics of those names that are gene/protein names. The criteria learned in this manner are then applied to our large set of names. We generated 193 classes of names and ILP led to criteria defining a select subset of 1,240,462 names. A simple false positive filter was applied to remove 8% of this set leaving 1,145,913 names. Examination of a random sample from this gene/protein name lexicon suggests it is composed of 82% (+/-3%) complete and accurate gene/protein names, 12% names related to genes/proteins (too generic, a valid name plus additional text, part of a valid name, etc.), and 6% names unrelated to genes/proteins. The lexicon is freely available at ftp.ncbi.nlm.nih.gov/pub/tanabe/Gene.Lexicon.

Algorithms↗

PubMed2004

Toward a mechanistic psychology of dialogue.

Traditional mechanistic accounts of language processing derive almost entirely from the study of monologue. Yet, the most natural and basic form of language use is dialogue. As a result, these accounts may only offer limited theories of the mechanisms that underlie language processing in general. We propose a mechanistic account of dialogue, the interactive alignment account, and use it to derive a number of predictions about basic language processes. The account assumes that, in dialogue, the linguistic representations employed by the interlocutors become aligned at many levels, as a result of a largely automatic process. This process greatly simplifies production and comprehension in dialogue. After considering the evidence for the interactive alignment model, we concentrate on three aspects of processing that follow from it. It makes use of a simple interactive inference mechanism, enables the development of local dialogue routines that greatly simplify language processing, and explains the origins of self-monitoring in production. We consider the need for a grammatical framework that is designed to deal with language in dialogue rather than monologue, and discuss a range of implications of the account.

Humans↗

PubMed1995

Issues in the structuring and acquisition of an ontology for medical language understanding.

Medical natural language understanding basically aims at representing the contents of medical texts in a formal, conceptual representation. The understanding process itself increasingly relies on a body of domain knowledge, generally expressed in the same conceptual formalism. The design of such a conceptual representation is a key knowledge-acquisition issue. When representing knowledge, the most important point is to ensure that the formal exploitation of the knowledge representation conforms to its meaning in the domain. We examined some methodological and theoretical principles to enforce this conformity. These principles result from our experience in MENELAS, a medical language understanding project.

Artificial Intelligence↗

PubMed2003

Automatic concept extraction from spoken medical reports.

OBJECTIVE: The objective of this project is to investigate methods whereby a combination of speech recognition and automated indexing methods substitute for current transcription and indexing practices. METHODS: We based our study on existing speech recognition software programs and on NOMINDEX, a tool that extracts MeSH concepts from medical text in natural language and that is mainly based on a French medical lexicon and on the UMLS. For each document, the process consists of three steps: (1) dictation and digital audio recording, (2) speech recognition, (3) automatic indexing. The evaluation consisted of a comparison between the set of concepts extracted by NOMINDEX after the speech recognition phase and the set of keywords manually extracted from the initial document. The method was evaluated on a set of 28 patient discharge summaries extracted from the MENELAS corpus in French, corresponding to in-patients admitted for coronarography. RESULTS: The overall precision was 73% and the overall recall was 90%. Indexing errors were mainly due to word sense ambiguity and abbreviations. A specific issue was the fact that the standard French translation of MeSH terms lacks diacritics. A preliminary evaluation of speech recognition tools showed that the rate of accurate recognition was higher than 98%. Only 3% of the indexing errors were generated by inadequate speech recognition. DISCUSSION: We discuss several areas to focus on to improve this prototype. However, the very low rate of indexing errors due to speech recognition errors highlights the potential benefits of combining speech recognition techniques and automatic indexing.

Abstracting and Indexing↗

PubMed1991

HF-Explain: a natural language generation system for explaining a medical expert system.

Causal models have been used, with considerable success, to reason in the medical domain. While these systems typically have a robust reasoning mechanism and knowledge base about their specific area of expertise, their ability to satisfactorily explain their results in a meaningful, coherent and concise manner has been less impressive then their diagnostic capabilities. This paper describes a program, HF-Explain, that generates natural language explanations of one such system--the Heart Failure Program. HF-Explain, is loosely based on work done by McKeown in the Text system, using augmented transition networks (ATN) as a formalism to guide the explanation process. The result is a coherent, concise, accurate and rich explanation of Heart Failure Programs' diagnostic hypotheses.

Cardiac Output, Low↗

PubMed2003

Text mining neuroscience journal articles to populate neuroscience databases.

We have developed a program NeuroText to populate the neuroscience databases in SenseLab (http://senselab.med.yale.edu/senselab) by mining the natural language text of neuroscience articles. NeuroText uses a two-step approach to identify relevant articles. The first step (pre-processing), aimed at 100% sensitivity, identifies abstracts containing database keywords. In the second step, potentially relevant abstracts identified in the first step are processed for specificity dictated by database architecture, and neuroscience, lexical and semantic contexts. NeuroText results were presented to the experts for validation using a dynamically generated interface that also allows expert-validated articles to be automatically deposited into the databases. Of the test set of 912 articles, 735 were rejected at the pre-processing step. For the remaining articles, the accuracy of predicting database-relevant articles was 85%. Twenty-two articles were erroneously identified. NeuroText deferred decisions on 29 articles to the expert. A comparison of NeuroText results versus the experts' analyses revealed that the program failed to correctly identify articles' relevance due to concepts that did not yet exist in the knowledgebase or due to vaguely presented information in the abstracts. NeuroText uses two "evolution" techniques (supervised and unsupervised) that play an important role in the continual improvement of the retrieval results. Software that uses the NeuroText approach can facilitate the creation of curated, special-interest, bibliography databases.

Abstracting and Indexing↗

PubMed1999

Automating a severity score guideline for community-acquired pneumonia employing medical language processing of discharge summaries.

Obtaining encoded variables is often a key obstacle to automating clinical guidelines. Frequently the pertinent information occurs as text in patient reports, but text is inadequate for the task. This paper describes a retrospective study that automates determination of severity classes for patients with community-acquired pneumonia (i.e. classifies patients into risk classes 1-5), a common and costly clinical problem. Most of the variables for the automated application were obtained by writing queries based on output generated by MedLEE1, a natural language processor that encodes clinical information in text. Comorbidities, vital signs, and symptoms from discharge summaries as well as information from chest x-ray reports were used. The results were very good because when compared with a reference standard obtained manually by an independent expert, the automated application demonstrated an accuracy, sensitivity, and specificity of 93%, 92%, and 93% respectively for processing discharge summaries, and 96%, 87%, and 98% respectively for chest x-rays. The accuracy for vital sign values was 85%, and the accuracy for determining the exact risk class was 80%. The remaining 20% that did not match exactly differed by only one class.

Community-Acquired Infections↗

PubMed1995

The roles of language processing in a spoken language interface.

This paper provides an overview of the colloquium's discussion session on natural language understanding, which followed presentations by M. Bates [Bates, M. (1995) Proc. Natl. Acad. Sci. USA 92, 9977-9982] and R. C. Moore [Moore, R. C. (1995) Proc. Natl. Acad. Sci. USA 92, 9983-9988]. The paper reviews the dual role of language processing in providing understanding of the spoken input and an additional source of constraint in the recognition process. To date, language processing has successfully provided understanding but has provided only limited (and computationally expensive) constraint. As a result, most current systems use a loosely coupled, unidirectional interface, such as N-best or a word network, with natural language constraints as a postprocess, to filter or resort the recognizer output. However, the level of discourse context provides significant constraint on what people can talk about and how things can be referred to; when the system becomes an active participant, it can influence this order. But sources of discourse constraint have not been extensively explored, in part because these effects can only be seen by studying systems in the context of their use in interactive problem solving. This paper argues that we need to study interactive systems to understand what kinds of applications are appropriate for the current state of technology and how the technology can move from the laboratory toward real applications.

Communication↗

PubMed2002

A critical period for right hemisphere recruitment in American Sign Language processing.

Signed languages such as American Sign Language (ASL) are natural languages that are formally similar to spoken languages, and thus present an opportunity to examine the effects of language structure and modality on the neural organization for language. Native learners of spoken languages show predominantly left-lateralized patterns of neural activation for language processing, whereas native learners of ASL show extensive right hemisphere (RH) and LH activation. We demonstrate that the RH angular gyrus is active during ASL processing only in native signers (hearing, ASL-English bilinguals) but not in those who acquired ASL after puberty (hearing, native English speakers). This is the first demonstration of a 'sensitive' or 'critical' period for language in an RH structure. This has implications for language acquisition and for understanding age-related changes in neuroplasticity more generally.

Adult↗

PubMed1995

Representing clinical narratives using conceptual graphs.

The analysis of medical narratives and the generation of natural language expressions are strongly dependent on the existence of an adequate representation language. Such a language has to be expressive enough in order to handle the complexity of human reasoning in the domain. Sowa's Conceptual Graphs (CG) are an answer, and this paper presents a multilingual implementation, using French, English and German. Current developments demonstrate the feasibility of an approach to natural Language Understanding where semantic aspects are dominant, in contrast to syntax driven methods. The basic idea is to aggregate blocks of words according to semantic compatibility rules, following a method called Proximity Processing. The CG representation is gradually built, starting from single words in a semantic lexicon, to finally give a complete representation of the sentence under the form of a single CG. The process is dependent on specific rules of the medical domain, and for this reason is largely controlled by the declarative knowledge of the medical Linguistic Knowledge Base.

Artificial Intelligence↗

PubMed2003

Working memory and lexical ambiguity resolution as revealed by ERPs: a difficult case for activation theories.

This series of three event-related potential experiments explored the issue of whether the underlying mechanism of working memory (WM) supporting language processing is inhibitory or activational in nature. These different cognitive mechanisms have been proposed to explain the more efficient processing of subjects with a high WM span compared to those with a low WM span. Participants with high and low WM span were presented with sentences containing a homonym followed three words later by a nominal disambiguation cue and a final disambiguation using a verb. At the position of the disambiguation cue, inhibitory or activational WM mechanisms predict contrasting results. When activation is the underlying mechanism for efficient processing, the prediction is that high memory span persons activate both meanings of the homonym equally in WM, whereas low memory span persons only have one meaning present. When inhibition is the underlying mechanism, the predictions are the reverse. The ERP data, in particular, the variations of the meaning related N400 component, showed clear evidence for inhibition as the underlying cognitive mechanism in high-span readers. For low-span participants the cueing towards the dominant or the subordinate meaning elicited an equivalently large N400 component suggesting that both meanings are active in WM. In high-span subjects, the dominant disambiguation cue elicited a smaller N400 than the subordinate one, indicating that for these subjects particularly the dominant meaning is active. The experiments showed that inhibitory processes are probably underlying WM used during language comprehension in high-span subjects. Moreover, they demonstrate that these subjects can use their inhibition in a more flexible manner than low-span subjects. The effects that these processing differences have on the efficiency of language parsing are discussed.

Adult↗

PubMed2001

Personalizing retrieval of journal articles for patient care.

We present a system for patient-specific searches on a database of medical journal articles which uses natural language techniques to match search results against patient records. We performed an information retrieval experiment comparing the performance of this system to two strategies, one of which uses extensive medical knowledge, while the other uses the same patient information our system has. The results show that our system is useful in improving recall over the strategy simulating a human specialist, and clearly outperforms the strategy of using the patient record content without intelligent processing.

Algorithms↗

PubMed2006

Linguistic complexity and information structure in Korean: evidence from eye-tracking during reading.

The nature of the memory processes that support language comprehension and the manner in which information packaging influences online sentence processing were investigated in three experiments that used eye-tracking during reading to measure the ease of understanding complex sentences in Korean. All three experiments examined reading of embedded complement sentences; the third experiment additionally examined reading of sentences with object-modifying, object-extracted relative clauses. In Korean, both of these structures place two NPs with nominative case marking early in the sentence, with the embedded and matrix verbs following later. The type (pronoun, name or description) of these two critical NPs was varied in the experiments. When the initial NPs were of the same type, comprehension was slowed after participants had read the sentence-final verbs, a finding that supports the view that working memory in language comprehension is constrained by similarity-based interference during the retrieval of information necessary to determine the syntactic or semantic relations between noun phrases and verb phrases. Ease of comprehension was also influenced by the association between type of NP and syntactic position, with the best performance being observed when more definite NPs (pronouns and names) were in a prominent syntactic position (e.g., matrix subject) and less definite NPs (descriptions) were in a non-prominent syntactic position (embedded subject). This pattern provides evidence that the interpretation of sentences is facilitated by consistent packaging of information in different linguistic elements.

Asian People↗

PubMed1991

SISCOPE: a multiuser information system for gastrointestinal endoscopy.

SISCOPE is an integrated data management system for use in gastrointestinal endoscopy units which operates in the multiuser mode on UNIX minicomputers or MS-DOS personal computers and can be used for patient bookings, endoscopic data entry and retrieval, and automatic report generation in upper gastrointestinal endoscopy, proctologic examinations, colonoscopy and peritoneoscopy. The description of endoscopic findings is remarkably detailed and data entry very rapid due to an advanced design of input screens that incorporates several recent concepts, including windows, menu bars and pull-down menus; typing is eliminated as data is entered with a mouse by pointing at options within menus. Endoscopic findings can be described under eight headings: morphology, topography, qualifiers, modifiers, signs of bleeding, endoscopic diagnosis, pathological diagnosis and etiology. Terminology is based strictly 3on the OMED system. SISCOPE also allows recording of details on endoscopic procedures, indications for the examination, preparation, premedication, complications and late entry of pathology reports. After entering all data, a report in natural language is produced automatically, the entire process taking one minute on average. Data retrieval programs give on-line access to previous examinations of a given patient and automatically generate activity reports. A formal language allows direct queries to the database and transfer of data for statistical analysis or other data processing. The system is simple to learn and use because operation is intuitive and all endoscopic techniques share the same basic menu structure and screen design.

Computer Systems↗

PubMed2005

Native and nonnative speakers' processing of a miniature version of Japanese as revealed by ERPs.

Several event-related potential (ERP) studies in second language (L2) processing have revealed a differential vulnerability of syntax-related ERP effects in contrast to purely semantic ERP effects. However, it is still debated to what extent a potential critical period for L2 acquisition, as opposed to the attained proficiency level in the L2, contributes to the pattern of results reported in previous ERP studies. We studied L2 processing within the model of a miniature version of a natural language, namely Japanese, specifically constructed to assure high proficiency of the learners. In an auditory ERP experiment, we investigated sentence processing of the "Mini-Japanese" in Japanese native speakers and German volunteers before and after training. By making use of three different types of violation, namely, word category, case, and classifier violations, native and nonnative ERP patterns were compared. The three types of violation elicited three characteristic ERP patterns in Japanese native speakers. The word category violation elicited an anteriorly focused, broadly distributed early negativity followed by a P600, whereas the case violation evoked a P600 which was preceded by an N400. The classifier violation led solely to a late left distributed negativity with an anterior focus. Although the P600 was similar for Japanese natives and learners, the N400 and the anterior negativities were not present in the learner group. The differences across groups suggest deviant neural processes in on-line syntactic and thematic processing in the L2 learners despite high behavioral skills.

Acoustic Stimulation↗

PubMed1995

Automated coding of patient discharge summaries using conceptual graphs.

In medicine, as in other domains, indexing and classification is a natural human task which is used for information retrieval and representation. In the medical field, encoding of patient discharge summaries is still a manual time-consuming task. This paper describes an automated coding system of patient discharge summaries from the field of coronary diseases into the ICD-9-CM classification. The system is developed in the context of the European AIM MENELAS project, a natural-language understanding system which uses the conceptual-graph formalism. Indexing is performed by using a two-step processing scheme; a first recognition stage is implemented by a matching procedure and a secondary selection stage is made according to the coding priorities. We show the general features of the necessary translation of the classification terms in the conceptual-graph model, and for the coding rules compliance. An advantage of the system is to provide an objective evaluation and assessment procedure for natural-language understanding.

Artificial Intelligence↗

PubMed1996

Development and evaluation of a computerized admission diagnoses encoding system.

Hospital information systems designed to support the needs of health care professionals include patient data entered using both freetext and precoded storage schemes. A major disadvantage of freetext storage schemes is that data captured in this format can only be presented as is to the user for review tasks. In the view of many health care scientists, natural language understanding systems capable of identifying, extracting, and encoding information contained in freetext data may provide the necessary tools to overcome this weakness. This paper describes the development and evaluation of a such a system designed to encode freetext admission diagnoses. This system combines both semantic and syntactic linguistic analysis techniques. Evaluation results demonstrate the overall performance of this system to be reasonable, accurately encoding approximately 76% of admission diagnoses. Inefficiencies are primarily due to the inability of this system to generate encodings in roughly 15% of test cases. When encodings are produced, however, accuracy equals that of the current manual coding method. With further modification, this application can partially automate the coding process.

Algorithms↗

PubMed1994

Users conceptual views on medical information databases.

As information databases we consider all the kinds of information repositories that are handled by computer systems. When querying very large information databases, the end-users are often faced with the problem to parse their questions efficiently into the query languages of the computer systems. Conceptual graphs were initially designed for natural language analysis and understanding. Due to their closeness to semantic networks, their expressiveness is powerful enough to be applied to knowledge representation and use by computer systems. This work demonstrates that conceptual graphs are a suitable means to model both the information in patient databases and the queries to these databases, and that operations on graphs can compute the pattern matching process needed to provide the answers. A prototype that exploits this model is presented. Experiments have been made with the material furnished by the Unified Medical Language System project (version 2, 1992) of the National Library of Medicine, USA.

Abstracting and Indexing↗