Methods Inf Med 1981; 20(01): 19-23
DOI: 10.1055/s-0038-1635290
Original Article
Schattauer GmbH

Computer-Assisted Bayesian Diagnosis of Connective Tissue Diseases[*]

COMPUTER-UNTERSTÜTZTE BAYESSCHE DIAGNOSE VON BINDEGEWEBSKRANK-HEITEN
T. R. Walsh
1   From the Information Science Group and Department of Medicine, School of Medicine, University of Missouri, Columbia, Missouri
,
D. A. B. Lindberg
1   From the Information Science Group and Department of Medicine, School of Medicine, University of Missouri, Columbia, Missouri
,
E. J. Klink Jr.
1   From the Information Science Group and Department of Medicine, School of Medicine, University of Missouri, Columbia, Missouri
,
G. C. Sharp
1   From the Information Science Group and Department of Medicine, School of Medicine, University of Missouri, Columbia, Missouri
› Author Affiliations
Further Information

Publication History

Publication Date:
14 February 2018 (online)

Bayes’ Theorem was applied interactively with a microcomputer to a rheumatology data base to provide a differential diagnosis of six primary connective tissue diseases. This diagnosis was evaluated in terms of the clinical diagnosis reached by a consensus of at least two of three rheumatologists. For each patient case, the diagnosis having the highest likelihood, after taking into account disease prevalence and combined symptom frequencies, was selected by the computer. The diagnostic model was trained by data from thoroughly studied clinical cases, for each of which 44 clinical and laboratory findings were collected according to a standard data base protocol. Each case received one of six diagnoses based on full clinical appraisal. Comparison of the diagnoses revealed 100 per cent agreement between the computer and the clinicians for the training population of 137 cases, and 94.4 per cent agreement for an independent, thoroughly studied test set of 89 cases. During testing, we observed the system had a genuine educational value for medical students and non-rheumatologists. In addition, there appeared to be a potential future use in facilitating actual clinical problem solving.

Mittels eines Microcomputers wurde des Bayessche Theorem interaktiv auf Rheuma-Meßwerte zur Differential-Diagnose von sechs primären Bindegewebskrankheiten angewandt. Diese Diagnose wurde dann mit der klinischen Diagnose verglichen, auf die sich mindestens zwei von drei Rheumatologen geeinigt hatten. In jedem Fall wurde vom Computer die unter Berücksichtigung der Prävalenz sowie der kombinierten Symptom-Häufigkeiten wahrscheinlichste Diagnose ausgewählt. Das diagnostische Modell beruht auf Angaben von gründlich durchuntersuchten klinischen Fällen; für jeden dieser Fälle wurden 44 klinische oder Laboratoriumsbefunde vermittels eines standardisierten Datenerfassungsbogens gesammelt. Nach gründlicher klinischer Bewertung wurde jedem Fall eine von sechs Diagnosen zugeteilt. Ein Diagnosenvergleich ergab hundertprozentige Übereinstimmung zwischen Computer und Ärzten für 137 Fälle des ersten Trainings-Samples und 94,4% Übereinstimmung für eine unabhängig davon zusammengestellte Gruppe von 89 weiteren Fällen. Während der Testphase zeigte sich, daß das Verfahren erzieherischen Wert für die Ausbildung von Medizin-Studenten und Nicht-Rheumatologen hatte. Weiterhin erscheint die zukünftige Verwendung des Verfahrens zur Erleichterung aktueller klinischer Problemlösungen möglich.

* This work was supported by the following grants: USPH5 T15 LM07006-05, from the National Library of Medicine, PHS/ DHEW, »Training Program in Medical Information Science«; USPHl P60 AM 20658-02; USPH 5 P50 HS02569-03, from the National Center for Health Services Research, OASH, and NIH Clinical Research Center RR-00-287-12.


 
  • References

  • 1 Couch R. D. Computer Diagnosis — Review and Comment. Path. Ann 1976; 11: 141-159.
  • 2 Dannenburg A, Shapiro A. R, Fries J. F. Enhancement of Clinical Predictive Ability by Computer Consultation. Meth. Inform. Med 1979; 18: 10-14.
  • 3 Degroot M. H. Probability and Statistics. Reading, Mass: Addison-Wesley; 1975
  • 4 Duda R. O, Hart P. E. Pattern Classification and Scene Analysis. New York: John Wiley and Sons; 1973
  • 5 Fisz M. Probability Theory and Mathematical Statistics. New York: John Wiley and Sons; 1963
  • 6 Gustafson D. H, Kestly J. J, Greist J. H, Jensen N. M. Initial Evaluation of Subjective Bayesian Diagnostic System. Health Serv. Res 1971; 6: 204-213.
  • 7 Jacquez J. A. (Edit.) Computer Diagnosis and Diagnostic Methods. Springfield, Ill.: C. C. Thomas; 1972
  • 8 Ledley R. S, Lusted L. B. Reasoning Foundation of Medical Diagnosis. Science 1959; 130: 9-21.
  • 9 Lindberg D. A. B, Sharp G. C, Kingsland L. C III , Weiss S. M, Hayes S. P, Ueno H, Hazelwood S. Computer-Based Rheumatology Consultant. Proceed. MEDINFO 80, Tokyo Amsterdam, The Netherlands: North Holland; 1980: 1311-15.
  • 10 Lusted L. B. Introduction to Medical Decision Making. Springfield, Ill.: C. C. Thomas; 1968
  • 11 Mcneil B. J, Keeler E, Adelstein S. J. Primer on Certain Elements of Medical Decision Making. New Engl. J. Med 1975; 293: 211-215.
  • 12 Mcshane D. J, Harlow A, Kraines R. G, Fries J. F. TOD: A Software System for the ARAMIS Data Bank. Computer 1979; 12: 34-40.
  • 13 Patton D. G. Introduction to Clinical Decision Making. Semin. Nucl. Med 1978; 8: 273-282.
  • 14 Rogers W, Ryack B, Mueller G. Computer-Aided Medical Diagnosis — Literature Review. Int. J. biomed. Corn-put 1979; 10: 267-289.
  • 15 Rosati R. A, Mcneer J. F, Starmer C. F, Mittler B. S, Morris J. J, Wallace A. G. A New Information System for Medical Practice. Arch. intern. Med 1975; 135: 1017-1024.
  • 16 Shapiro A. R. The Evaluation of Clinical Predictions — A Method and Initial Application. New Engl. J. Med 296 1977 1509-1514.
  • 17 Szolovirs P, Pauker S. G. Categorical and Probabilistic Reasoning in Medical Diagnosis. Artif. Intell 1978; 11: 115-144.
  • 18 Wagner G, Tautu P, Wolber U. Problems of Medical Diagnosis. A Bibliography. Meth. Inform. Med 1978; 17: 55-74.
  • 19 Wardle D. J, Wardle L. Computer Aided Diagnosis — A Review of Research. Meth. Inform. Med 1978; 17: 15-28.