An Automated Parsing Routine for Diagnostic Statements of Surgical Pathology Reports

Abstract

An automated parsing routine was written for extracting the »site«, »diagnostic«, and »modifier« components of the diagnostic statements of the diagnostic summary of surgical pathology reports. Such parsed reports appear to be suitable for input into an information retrieval system for the surgical pathology reports.Data was input through a key-to-tape device producing a computer compatible magnetic tape with a record size of 870 bytes. The statements were parsed through syntactic and morphological analysis utilizing the common prepositions, the common punctuations and the morphemal constructions common in medical terms. (A total of sixty-two delimiters were used). Certain suffix transformations were performed, converting some »site« adjective to »site« nouns, and some »diagnostic« nouns to »site« nouns. 1,108 diagnostic statements were processed with an error rate of 9.3% for the latest version on the last 493 statements.