Examination of Disease Names Using Non-Abelian Symbolic Logic^[*]

Summary

Despite many investigations of the potential role of formal symbolic logic in medicine, little attention has been paid to the distribution of words in large quantities of natural language text and how this might affect the choice of a logic system. With the emergence of communicating word processor technology, it is now possible to enter large amounts of routine medical text in computer-readable form and study its implications for symbolic logic formalisms.

We examined the word distribution in the Anatomical Diagnosis reports of 7,000 consecutive autopsied patients from the Department of Pathology of The Johns Hopkins Hospital. We found that uncommon disease entities consisting of common component words were best recovered using a search algorithm which examines conjunction (and) logic in ascending order of word frequency and alternation (inclusive or) logic in descending order of word frequency. This logic is non-Abelian, non-distributive, and associative for conjunction but not for alternation. Our results suggest that natural language text analysis, which is essential for studying medical documents by computer, may require heretofore little studied variant forms of symbolic logic.

Trotz vieler Untersuchungen über die potentielle Rolle der formalen symbolischen Logik in der Medizin wurde bisher der Wortverteilung in umfangreichen Klartexten und ihrer Wirkung auf die Wahl des anzuwendenden logischen Systems wenig Aufmerksamkeit gewidmet. Mit dem Aufkommen der Wortprozessor/Kommunikationstechnologie ist es jetzt möglich, umfangreiche medizinische Texte in computerlesbarer Form einzuspeichern und die aus quantitativen Klartextanalysen gewonnenen Erkenntnisse für die Ausarbeitung des symbolisch-logischen Formalismus zu verwenden.

Wir untersuchten die Wortverteilung in den Sektionsberichten von 7000 aufeinanderfolgenden Obduktionen des Pathologischen Instituts des Johns Hopkins Hospital. Dabei fanden wir, daß ungewöhnliche, aus häufig gebrauchten Wortkombinationen bestehende Krankheitsbegriffe am besten wiedergefunden werden können, wenn der Suchalgorithmus die Konjunktionslogik (logisches »und«) in aufsteigender Reihenfolge und die Alternationslogik (logisches »oder«) in absteigender Reihenfolge der Worthäufigkeit enthält. Diese Logik ist nicht-Abelsch und nicht-distributiv. Sie ist assoziativ für die Konjunktion, aber nicht für die Alternation. Unsere Ergebnisse weisen darauf hin, daß zur Klartextanalyse, die einen wesentlichen Bestandteil der Computerverarbeitung medizinischer Texte darstellt, bisher wenig benutzte Formen der symbolischen Logik mit herangezogen werden sollten.

Key-Words

Natural Language Text Analysis - Symbolic Logic - Autopsy Protocol

Schlüssel-Wörter

Analyse der Umgangssprache - symbolische Logik - Sektionsbericht

• REFERENCES

• 1 Aller R. D. Computers in anatomic pathology. Clin. Lab. Med 1983; 3: 133-147.
• 2 Aller R. D, Robboy S. J, Poitras J. W, Altshuler B. S, Cameron M, Prior M. C, Miao S, Barnett G. O. Computer-assisted pathology encoding and reporting system (CAPER). An on-line computer system developed at the Massachusetts General Hospital. Amer. J. clin. Path 1977; 68: 715-720.
• 3 Barnett G. O, Justice N. S, Somand M. E, Adams J. B, Waxman B. D, Bea-man P. D, Parent M. S, VanDeusen E.R, Greenlie J. K. COSTAR- a computer-based medical information system for ambulatory care. Proceed. I.E.E.E 1979; 67: 1226-1237.
• 4 Bowie J, Barnett G. O. MUMPS -An economical and efficient time-sharing system for information management. Corn-put. Progr. Biomed 1976; 6: 11-22.
• 5 Card W. I, Good I. J. Logical foundations of medicine. Brit. med. J 1971; I: 718-720.
• 6 Chomsky N. On the nature of language. Ann. N. Y. Acad. Sci 1976; 280: 46-57.
• 7 Coles E. C, Slavin G. An evaluation of automatic coding of surgical pathology reports. J. clin. Path 1976; 29: 621-625.
• 8 Coté R. A, Robboy S. Progress in medical- information management. Systematized nomenclature of medicine (SNOMED). J. Amer. med. Ass 1980; 243: 756-762.
• 9 Dreyfus H. L. What Computers Can’t Do. A Critique of Artificial Reason. New York: Harper & Row; 1972
• 10 Foulis P. R, Norbut A. M, Mendelow H, Kessler G. F. Pathology accessioning and retrieval system with encoding by computer (PARSEC). Amer. J. clin. Path 1980; 73: 748-753.
• 11 Hirschman L, Story G, Marsh E, Lyman M, Sager N. An experiment in automated health care evaluation from narrative medical records. Comput. biomed. Res 1981; 14: 447-463.
• 12 Kayser K. Logic and diagnosis. Meth. Inform. Med 1975; 14: 76-80.
• 13 Knuth D. Sorting and Searching. Volume 3 in: The Art of Computer Programming. Reading/ Mass: Addison-Wesley; 1983: 106-111. 552-559
• 14 Ledley R. S, Lusted L. B. Reasoning foundations of medical diagnosis. Symbolic logic, probability, and value theory aid our understanding of how physicians reason. Science 1959; 130: 9-21.
• 15 Marcus M. P. Theory of Syntactic Recognition for Natural Language. In Winston P. H, Brown R. D. (Eds) Artificial Intelligence: An MIT Perspective. Cambridge/Mass: The MIT Press; 1979: 193-229.
• 16 Miller R. A, Pople Jr H. E, Myers J. D. Internist-I, an experimental computer-based diagnostic consultant for general internal medicine. New. Engl. J. Med 1982; 307: 468-476.
• 17 Miller R. E, Steinbach G. L, Dayhoff R. E. A hierarchical computer network: An alternative approach to clinical laboratory computerization in a large hospital. Proceed. 4th Symp. Comput. Ap-plic. Med. Care 1980; 1: 505-513.
• 18 Moore G. W, Haupt H. M, Hutchins G. M. A hypothesis test for causal explanations in human pathology: Evaluation of pulmonary edema in 181 autopsied patients with leukemia. Math. Biosc. 62. 1982: 253-279.
• 19 Moore G. W, Hutchins G. M. Symbolic logic analysis of cause of death in humans: Application to 108 patients after coronary artery bypass graft surgery. J. theor. Biol 1981; 92: 267-291.
• 20 Moore G. W, Hutchins G. M, Bulkley B. H. Certainty levels in the nullity method of symbolic logic: Application to the pathogenesis of congenital heart malformations. J. theor. Biol 1979; 76: 53-81.
• 21 Moore G. W, Hutchins G. M, Miller R. E. Strategies for searching medical natural language text: Distribution of words in the anatomical diagnoses of 7000 autopsied patients. Amer. J. Path 1984; 115: 36-41.
• 22 Paplanus S. H, Shepard R. H, Zvargulis J. E. A computer-based system for autopsy diagnosis storage and retrieval without numerical coding. Lab. Invest 1969; 20: 139-146.
• 23 Permuted Title Index. Fed. Proceed 1983; 42: 1421-1615.
• 24 Robboy S. J, Altshuler B. S, Chen H. Y. Retrieval in a computer-assisted pathology encoding and reporting system (CAPER). Amer. J. clin. Path 1981; 75: 654-661.
• 25 Röttger P, Reul H, Klein I, Sunkel H. Die vollautomatische Dokumentation und statistische Auswertung pathologischanatomischer Befundberichte. Meth. Inform. Med 1969; 8: 19-26.
• 26 Röttger P, Reul H, Sunkel H, Klein I. Neue Auswertungsmöglichkeiten pathologisch-anatomischer Befundberichte. Klartextanalyse durch Elektronenrechner. Meth. Inform. Med 1970; 9: 35-44.
• 27 Shortliffe E. H, Buchanan B. G, Feigenbaum E. A. Knowledge engineering for medical decision making: A review of computer-based clinical decision aids. Proc. I.E.E.E 1979; 67: 1207-1224.
• 28 Stein M. A, Winter J. Theory development in medical decision-making. Int. J. biomed. Comput 1974; 5: 147-159.
• 29 Story G, Hirschman L. Data base design for natural language medical data. J. med. Syst 1982; 6: 77-88.
• 30 Watanabe T, Ohsawa T, Suziki T. A simple database language for personal computers. Commun. ACM 1983; 26: 646-653.
• 31 Wong R. L, Reno J. D, Hain T. C, Platt R. C, Gaynon P. S, Joseph D. M. Profile of a dictionary compiled from scanning over one million words of surgical pathology narrative text. Comput. biomed. Res 1980; 13: 382-398.
• 32 Zeman J. J. Modal Logic. The Lewis-Modal Systems. London: Clarendon Press; 1973