A Method for Automatic Coding of Medical Information in Patient Records

An automated method for coding hospital patient records is described, which uses a combination of simple computer programs and a structured dictionary. The system accepts unrestricted medical language in French as input, and generates code combinations which represent the semantic values of the original statements. The coding structure is the trial version of the Systematized Nomenclature of Medicine (SNOMED) [9], A prototype system has been developed from a 20,000 word corpus of obstetrical and gynecological data, recorded from patient records written by French-speaking Quebec physicians. Preliminary test results on a small sample show that the system generates accurate codes and code combinations for 97.9% of the data tested. The programs were written in Fortran IV and implemented on a DEC PDP-9 computer. The dictionary keys are word segments and the entries contain operators, word segments and SNOMED codes. The dictionary structure permits a single entry to generate multiple SNOMED codes and to handle the types of linguistic paraphrase found in hospital records. The SNOMED codes and code combinations constitute a metalanguage which provides the means for data compression and uniform representation of natural language medical data.

Es wird eine automatisierte Methode zur Verschlüsselung von Krankenblättern beschrieben, die mit einer Kombination von einfachen Computerprogrammen und einem strukturierten Wörterbuch arbeitet. Das System nimmt uneingeschränkte Medizinsprache in Französisch als Eingabe an und schafft Schlüsselkombinationen, die die semantischen Werte der ursprünglichen Aussagen darstellen. Die Verschlüsselungsstruktur ist die Probefassung von Systematized Nomenclature of Medicine (SNOMED) [9]. Aus einem Corpus von 20.000 Wörtern geburtshilflicher und gynäkologischer Daten ist ein Prototypsystem entwickelt worden, entnommen aus Krankenblättern, die von französischsprachigen Ärzten in Quebec geschrieben wurden. Vorläufige Testergebnisse an einer kleinen Stichprobe zeigen, daß das System exakte Codes und Schlüsselkombinationen für 97,9% der getesteten Daten hervorbringt. Die Programme wurden in Fortran IV geschrieben und auf einem DEC PDP-9 Computer ausgeführt. Die Wörterbuchschlüssel sind Wortsegmente, und die Eintragungen enthalten Operatoren, Wortsegmente und SNOMED-Schlüssel. Die Wörterbuchstruktur ermöglicht es, daß eine einzige Eintragung vielfache SNOMED-Schlüssel hervorbringt und daß die in Krankenblättern auftauchenden Arten linguistischer Paraphrasen gehandhabt werden können. Die SNOMED-Schlüssel und Schlüsselkombinationen stellen eine Metasprache dar, welche die Mittel zur Datenkomprimierung und gleichförmigen Darstellung medizinischer Daten in natürlicher Sprache liefert.

Key-Words

Automatic Coding - Natural Language Text Processing - Computer Dictionary Structures - Word Segmentation Methods - French Medical Statements - SNOMED

Schlüssel-Wörter

Automatische Verschlüsselung - Textverarbeitung in natürlicher Sprache - Computerwörterbuch-Strukturen - Methoden der Wortsegmentierung - französische medizinische Aussagen - SNOMED

• References

• 1 Automatic – Processing Advisory Committee. Language and Machines: Computers in Translation and Linguistics. Washington/D. C: National Academy of Sciences/National Research Council, Publication; 1416 1966
• 2 Babkman B, Bernier L, Cousineau L, Tanguay G. The Translation of SNOP: A First Step in the Construction of an Automated Medical Lexicon. Meta 19 1974; 28-42.
• 3 Brothers D. A. Assigning ICD Codes by Computer. Working Paper. (Special Projects Section, Health Division, Statistics Canada, November, 1974.)
• 4 Chomsky N. Recent Contributions to the Theory of Innate Ideas. Synthese 17 1967; 2-11.
• 5 Chomsky N. Syntactic Structures. The Hague; Mouton: 1957
• 6 Chomsky N. Aspects of the Theory of Syntax. Cambridge/Mass: MIT Press; 1965
• 7 College of American Pathologists. Systematized Nomenclature of Pathology. Chicago/Ill: 1965
• 8 Commission on Professional and Hospital Activities. Hospital Adaptation of ICDA. 2nd Edition).. Ann Arbor/Mich; 1973
• 9 Coté R. A. (Editor-in-Chief): Systematized Nomenclature of Medicine (Trial Version). Chicago/Ill: College of American Pathologists; 1975
• 10 Coté R. A. A Systematized Nomenclature of Medicine for the Health Care System. Proceedings of the 4th Annual Conference of the Society for Computer Medicine, New Orleans/La., November 1974. Section 4.8, 1–12.
• 11 Derwing B. L. Transformational Grammar as a Theory of – Acquisition: A Study in the Empirical, Conceptual and Methodological Foundations of Contemporary Linguistic Theory. Cambridge: Cambridge University Press; 1973
• 12 Dinwoodie H. P, Howell R. W. Automatic Disease Coding, the »Fruit Machine« Method in General Practice. Brit. J. prevent, soc. Med 27 1973; 59-63.
• 13 Dunham G. S. The NIH Pathology Diagnoses Encoder. Internal Report. Division of Computer Research and Technology, National Institutes of Health; Bethesda/Md: 1975
• 14 Federal Bureau of Statistics. International Classification of Diseases Adapted (8th Revision). Ottawa: 1968
• 15 Graitson M. Identification et transformation automatiques des morphèmes terminaux dans le lexique medical franyais. Cahiers de Lexieologie 26 1975; 85-109.
• 16 Greenwood R. M. KODIAC. A System for Disease Coding by a Medium-Sized Computer. Int. J. bio-med. Comput 03 1972; 123-134.
• 17 Harris Z. Methods in Structural Linguistics. Chicago/Ill: University of Chicago Press; 1951
• 18 Katz J. J, Fodor J. A. The Structure of a Semantic Theory. 39 1963; 170-210.
• 19 Lamson B. G, Gunski B. C, Hawthorne G. S, Soutter J. C, Russell W. S. Lamson-IBM Thesaurus. Proceedings of the 7th IBM Medical Symposium. Poughkeepsie/N.Y: 1965
• 20 Manuila A, Manuila L, Nicole M, Lambert H. (Eds): Dictionnaire franq-ais de médecine et de biologie. 4 Vol. Paris: Masson et Cie; 1970-1975
• 21 Montgomery C. A. Linguistics and Information Science. J. Amer. Soc. Inform. Sei 02 1972; 195-2X9.
• 22 Morgan J. D. A Computerized »Conversational« Technique to Form Numerically Coded Medical Diagnoses. University of Utah Ph. D. Dissertation. Salt Lake City/Utah: 1971
• 23 Otten M, Pacak M. Intermediate –s for Automatic – Processing. Software Engineering. (Vol. 3). New York and London: Academic Press; 1970
• 24 Pacak M. Computational Linguistics and Information Handling. In Howerton P. W. (Edit.): Management of Information Handling Systems. Rochelle Park/N.J: Hayden Book Co., Inc; 1974
• 25 Pacak M. G, Pratt A. W, White W. C. Automated Morphosyntactic Analysis of Medical –. Inform. Process. Manag 12 1976; 71-76.
• 26 Pratt A. W. Medicine, Computers, and Linguistics. Advances in Biomedical Engineering. (Vol. 3). New York: Academic Press; 1973
• 27 Pratt A. W. Automatic Processing of Pathology Data. Proceedings of IRIA Symposium on Medical Data Processing; Toulouse: March 1971
• 28 Pratt A. W, Pacak M. Identification and Transformation of Terminal Morphemes in Medical English. Meth. Inform. Med 05 1969; 84-90.
• 29 Salton G. Automatic Information Organization and Retrieval. New York: McGraw-Hill; 1968
• 30 Sapir E. Language. New York: Harcourt, Brace and Co; 1921
• 31 Thompson E. T, Hayden A. C. (Eds): Standard Nomenclature of Diseases and Operations. (5th Edition).. New York, Toronto and London: McGraw-Hill; 1961
• 32 World Health Organization. International Classification of Diseases. (8th Revision).. Geneva: 1967