Use of general-purpose negation detection to augment concept indexing of medical documents: a quantitative study using the UMLS.

OBJECTIVES: To test the hypothesis that most instances of negated concepts in dictated medical documents can be detected by a strategy that relies on tools developed for the parsing of formal (computer) languages-specifically, a lexical scanner ("lexer") that uses regular expressions to generate a finite state machine, and a parser that relies on a restricted subset of context-free grammars, known as LALR(1) grammars. METHODS: A diverse training set of 40 medical documents from a variety of specialties was manually inspected and used to develop a program (Negfinder) that contained rules to recognize a large set of negated patterns occurring in the text. Negfinder's lexer and parser were developed using tools normally used to generate programming language compilers. The input to Negfinder consisted of medical narrative that was preprocessed to recognize UMLS concepts: the text of a recognized concept had been replaced with a coded representation that included its UMLS concept ID. The program generated an index with one entry per instance of a concept in the document, where the presence or absence of negation of that concept was recorded. This information was used to mark up the text of each document by color-coding it to make it easier to inspect. The parser was then evaluated in two ways: 1) a test set of 60 documents (30 discharge summaries, 30 surgical notes) marked-up by Negfinder was inspected visually to quantify false-positive and false-negative results; and 2) a different test set of 10 documents was independently examined for negatives by a human observer and by Negfinder, and the results were compared. RESULTS: In the first evaluation using marked-up documents, 8,358 instances of UMLS concepts were detected in the 60 documents, of which 544 were negations detected by the program and verified by human observation (true-positive results, or TPs). Thirteen instances were wrongly flagged as negated (false-positive results, or FPs), and the program missed 27 instances of negation (false-negative results, or FNs), yielding a sensitivity of 95.3 percent and a specificity of 97.7 percent. In the second evaluation using independent negation detection, 1,869 concepts were detected in 10 documents, with 135 TPs, 12 FPs, and 6 FNs, yielding a sensitivity of 95.7 percent and a specificity of 91.8 percent. One of the words "no," "denies/denied," "not," or "without" was present in 92.5 percent of all negations. CONCLUSIONS: Negation of most concepts in medical narrative can be reliably detected by a simple strategy. The reliability of detection depends on several factors, the most important being the accuracy of concept matching.

Voice-activated retrieval of mammography reference images.

We undertook this project to integrate context sensitive computer-based educational and decision making aids into the film interpretation and reporting process, and to determine the clinical utility of this method as a guide for further system development. An image database of 347 digital mammography images was assembled and image features were coded. An interface was developed to a computerized speech recognition radiology reporting system which was modified to translate reported findings into database search terms. These observations were used to formulate database search strategies which not only retrieved similar cases from the image database, but also other cases that were related to the index case in different ways. The search results were organized into image sets intended to address common questions that arise during image interpretation. An evaluation of the clinical utility of this method was performed as a guide for further system development. We found that voice dictation of prototypical mammographic cases resulted in automatic retrieval of reference images. The retrieved images were organized into sets matching findings, diagnostic hypotheses, diagnosis, spectrum of findings or diagnoses, closest match to dictated case, or user specified parameters. Two mammographers graded the clinical utility of each form of system output. We concluded that case specific and problem specific image sets may be automatically generated from spoken case dictation. A potentially large number of retrieved images may be divided into subsets which anticipate common clinical problems. This automatic method of context sensitive image retrieval may provide a "continuous" form of education integrated into routine case interpretation.

The prospect of expert system-based cognitive support as a by-product of image acquisition and reporting.

In order for computer-based decision-support tools to find routine use in the everyday practice of clinical radiology, further development of user interface and knowledge content are required. In an ideal interface, the interaction between the radiologist and the computer would be minimized and painlessly integrated into existing work patterns. In this article, we explore some of the ways that pre-existing computer interactions in the processes of image acquisition and reporting can be used to feed case information into an expert system and thereby allow users to acquire advice from it in an automatic fashion. We describe interface models that we have developed in the domains of mammography and obstetric ultrasound, and discuss interface and content-related questions that have arisen from informal evaluations of these systems. In particular, the need for clinical outcome-relevant decision support and training level-appropriate decision support are discussed in detail.

Knowledge-based radiologic image retrieval using axes of clinical relevance.

This paper describes an approach to computer-based intelligent retrieval of feature-coded radiographic images relevant to a specific case being evaluated. The approach involves partitioning the search space along clinically natural groups of attributes which we call "axes of clinical relevance." By embedding knowledge about the domain to help direct the search process, a clinician's needs may be met more comprehensively. Domain knowledge, supplied to the system as "axis heuristics," may make search more robust. These heuristics provide a graded, progressive relaxation of the search constraints. This approach helps show the user groups of images in order of probable relevance to a current case. AXON is a prototype knowledge-based system constructed to illustrate this approach in the domain of chest imaging. This paper describes the AXON system, demonstrates some searches which illustrate the potential utility of this approach, and discusses preliminary tests of the search strategies used by AXON.

Expert system-controlled image display.

Conventional computer-based medical expert systems deliver advice to physicians as written text. While such advice is useful, it has distinct limitations in a visually oriented discipline such as diagnostic radiology, in which decisions often depend on pattern recognition and appreciation of subtle morphologic features. The authors developed a prototype expert computer system, IMAGE/ICON, which displays groups of images sorted into a series of axes based on different ways in which they may be similar. They may share a common feature, group of features, causes, or clinical setting. IMAGE/ICON may display examples of morphologic variations of a dominant finding or a spectrum of abnormalities seen in an specific disease or group of diseases. The system also assembles a written analysis of key features of a case. Such a tool may be useful as a diagnostic aid or for continuing medical education. It is likely to have particular impact in the form of an intelligent radiologic workstation, as picture archiving and communication systems become available.

Behavioral scoring using programmable calculators: a technique usable in the field.

A simple method of recording the time spent in various behavioral categories during behavioral scoring is described. Use is made of a programmable calculator which is made to function as a multiple timer, keeping track of each of the categories. Any number of mutually exclusive categories can be scored using a single key press, by assigning a pre-set code to each. A print-out of the analysed frequency or duration data can be obtained either concurrently or at any time after the experiment, as required. The least count of the technique is about 1-2 seconds and this precludes its use for extremely rapidly changing behaviors. Apart from this, it is convenient, time-saving and especially suitable for field use.