SAGER ET AL., Natural Language Processing Natural Language Processing and the Representation. of Clinical Data

Abstract 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. n J Am Med Informatics Assoc. 1994;1:142-160. The attention currently being given to the computer-based patient record 1,2 adds impetus to the devel-opment of representational structures for ‘clinical data. It is hoped that standardized structures can be de-veloped to serve as a framework for combining pa-tient data from multiple sources. In the evolution of computerized patient record systems, the contro-versy between free text and preset categories for re-cording patient data has not been resolved. The need for standards pushes toward preset categories and controlled vocabularies, while the need for expressive power, so as not to distort the patient data, speaks for allowing some amount of free-text reporting. A compromise that is not compromising is called for. It is the aim of this paper to show that the techniques of linguistic analysis and natural-language processing (NLP) can contribute to this effort. Within medical informatics there has been a long-standing concern with medical language. In a 197