Evaluation of automatic and manual knowledge acquisition for cerebrospinal fluid (CSF) diagnosis

  • A. Ultsch
  • T. O. Kleine
  • D. Korus
  • S. Farsch
  • G. Guimarães
  • W. Pietzuch
  • J. Simon
Knowledge Acquisition and Learning
Part of the Lecture Notes in Computer Science book series (LNCS, volume 1211)


Neural Network technology for knowledge acquisition appears to be promising. Neural Networks alone, however, are insufficient for a medical domain, because of their inability to explain decisions. Knowledge conversion procedures from neural networks to symbolic knowledge are necessary. As a first milestone of the MEDWIS project NELA a knowledge conversion algorithm, called sig*, was tested. We compared a knowledge base generated by sig* with a first prototype of a manually built knowledge base for the diagnosis of cerebrospinal fluid. We have found the performance of the automatically build knowledge base to be quantitatively comparable to the manually acquired knowledge base. The quality of the extracted rules with respect to human understanding and medical plausibility was evaluated by use of a questionnaire. A low number of missing parameters found together with a correct choice of parameters in most cases indicated that the extracted rules correspond well to knowledge expressed by a medical expert. The results presented here confirm that a part of our automatic knowledge acquisition procedure, that is conversion of knowledge from subsymbolic representation to a rule based representation of knowledge, performed comparably, proved to be superior to manual knowledge acquisition with some respects.


Knowledge Base Knowledge Acquisition Laboratory Diagnosis Acute Bacterial Meningitis Symbolic Knowledge 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [1]
    Beach, S. S., Gevarter, W.: Standarts for Evaluating Expert System Tools. in: Expert Systems with Applications, Vol. 2, 1991, 259–267.Google Scholar
  2. [2]
    Bense, H., Bodrow, W. (Eds.) (1995) Objektorientierte und regelbasierte Wissensverarbeitung, Spektrum Akademischer Verlag GmbH, Heidelberg.Google Scholar
  3. [3]
    Berlit, P (1992) Klinische Neurologie. VCH Weinheim.Google Scholar
  4. [4]
    Deichsel, G., Trampisch, H.J.: Clusteranalyse and Diskriminanzanalyse. Gustav Fisher Verlag, Stuttgart, 1985.Google Scholar
  5. [5]
    Engelbrecht, R., Rector, A., Moser, W.: Verification and Validation. in: Assessment and Evaluation of Information Technologies, E.M.S-J-van Gennip and J.L. Talmon (eds.), IOS Press, 1995, 51–66.Google Scholar
  6. [6]
    Felgenhauer, K. (1988, 1992) Liquordiagnostik. In: L. Thomas: Labor und Diagnose. S. 1403–1423. Medizinische Verlagsgesellschaft Marburg, 1403–1423.Google Scholar
  7. [7]
    Gallant, S.I.: Neural Network Learning and Expert Systems, MIT Press, Cambridge, MA, 1993.Google Scholar
  8. [8]
    Hartung, J.: Statistik, 8. Aufl., Oldenbourg Verlag, 1991.Google Scholar
  9. [9]
    Kleine, T. O. (1979) New developments in the diagnosis of cerebrospinal fluid. Report on the Workshop Conference of the German Society for Clinical Chemistry. J. Clin. Chem. Clin. Biochem. 17, 505–511Google Scholar
  10. [10]
    Kleine, T. O.: (1989) New diagnostic methods for inflammation of the human central nervous system by cerebrospinal fluid analysis. Workshop Conference of the German Society for Clinical Chemistry. J. Clin. Chem. Clin. Biochem. 27, 895–932Google Scholar
  11. [11]
    Kleine, T.O. (1980) Neue Labormethoden für die Liquordiagnostik. Thieme, Stuttgart 1980Google Scholar
  12. [12]
    Kleine, T.O. (1984): Diagnostische Untersuchungen im Liquor cerebrospinalis. In: L. Thomas: Labor und Diagnose. S.937–965. Medizinische Verlagsgesellschaft MarburgGoogle Scholar
  13. [13]
    Kleine, T.O. (1989): Nervensystem. In: H.Greiling, A.M. Gressner: Lehrbuch der Klinischen Chemie und Pathobiochemie, Schattauer Stuttgart, 859–893.Google Scholar
  14. [14]
    Kleine, T.O. (1991): D-Lactat und L-Lactat im Liquor cerebrospinalis bei akuten entzündlichen Erkrankungen des Zentralnervensystems. Lab. med. 15, (114–116.Google Scholar
  15. [15]
    Kleine, T.O., Baerlocher, K., Niederer, V., Keller, H., Reutter, F., Tritschler, W., Bablok, W. (1979) Diagnostische Bedeutung der Lactatbestimmung im Liquor der Minigitis, Dtsch. Med. Wschr. 104(15), 553–557Google Scholar
  16. [16]
    Kleine, T.O., Hackler, R., Lehmitz, R., Meyer-Rienecker, H. (1994): Liquor-diagnostik: Klinisch-chemische Kenngrößen — eine kritische Bilanz. DG Klinische Chemie Mitteilungen 25, 199–214.Google Scholar
  17. [17]
    Kleine, T.O., Hackler, R., Lütcke, A., Zöfel, P., Albrecht, J., Rumpl, E., Meyer-Rienecker, H.J. (1994) Laboratory diagnosis of radicular and pseudoradicular Syndromes in Cerebrospinal Fluid (CSF): Reliability of methods in consideration of pathogenetic aspects, Eur.J.Clin.Chem.Clin.Biochem. 32, 45–52.Google Scholar
  18. [18]
    Kleine, T.O., Weber, L., Zöfel, P. (1987) Diagnostische Aussage eines Basis-Liquor-Programmes bei Erkrankungen des Zentralnervensystems (ZNS), Lab. med. 11, 175–176Google Scholar
  19. [19]
    Kleine, T.O., Wittmann, M. (1995): Protein determination in urine and cerebrospinal fluid (CSF) with the pyrogallol red molybdate method compared with the biuret method. Abstract SEPaC Congress Vienna 1995. Wien. Klin. Wschr. 107 Suppl. 199, 10.Google Scholar
  20. [20]
    Kohonen, T.: Self-Organizing Maps. Springer, Berlin, 1995.Google Scholar
  21. [21]
    Lehmitz, R., Kleine, T.O. (1994): Liquorzytologie: Ausbeute, Verteilung und Darstellung von Leukozyten bei drei Sedimentationsverfahren im Vergleich zu drei Zytozentrifugations-Modifikationen. Lab. med. 18, 91–99.Google Scholar
  22. [22]
    Michaelis, J., Wellek, S., Willems, J.L. (1990) Reference Standards for Software Evaluation, Methods of Information in Medicine 29, F. K. Schattauer Verlagsgesellschaft mbH, 289–297.Google Scholar
  23. [23]
    Ohmann, C., Belenky, G.: Leitfaden zur Evaluierung von Wissensbasen. MEDWIS-Arbeitskreis Evaluation, 1996.Google Scholar
  24. [24]
    Puppe, F. (1991) Einführung in Expertensysteme, 2. Auflage, Springer Verlag, Berlin.Google Scholar
  25. [25]
    Quinlan, J.R. (1984) Learning Efficient Classification Procedures and their Application to Chess End Games, in: Michalski, R.; Carbonell, J.G.; Mitchell, T.M. (Eds.) Machine Learning — An Artificial Intelligence Approach, Berlin, 463–482.Google Scholar
  26. [26]
    Sax, L. (1992) Angewandte Statistik, 7. Auflage, Springer-Verlag, Berlin.Google Scholar
  27. [27]
    Ultsch, A, Korus, D., Kleine, T.O. (1995) Integration of Neural Networks and Knowledge-Based Systems in Medicine, 5th. Conf. Artificial Intelligence in Medicine Europe AIME'95, Pavia, Italy, June 1995, in: Barahona, P.; Stefanelli, M.; Wyatt, J. Artificial Intelligence in Medicine, Lecture Notes in Artificial Intelligence 934, Springer, 425–426.Google Scholar
  28. [28]
    Ultsch, A. (1993) Knowledge Extraction from Self-organizing Neural Networks, in O. Opitz, B. Lausen and R. Klar, (Eds.) Information and Classification, Berlin: Springer, 301–306.Google Scholar
  29. [29]
    Ultsch, A. (1993) Self-Organized Feature Maps for Monitoring and Knowledge Acquisition of a Chemical Process, in Gielen, S., Kappen, B. (Eds.): Proc. Intl. Conf. on Artificial Neural Networks (ICANN), Amsterdam, Sep. 1993, Springer, 864–867.Google Scholar
  30. [30]
    Ultsch, A. (1993) Self-organizing Neural Networks for Visualization and Classification, in O. Opitz, B. Lausen and R. Klar, (Eds.) Information and Classification, Springer-Verlag, Berlin 307–313.Google Scholar
  31. [31]
    Ultsch, A. (1994) The Integration of Neural Networks with Symbolic Knowledge Processing, in: Diday et al.: New Approaches in Classification and Data Analysis, Springer, 445–454.Google Scholar
  32. [32]
    Ultsch, A., Korus, D. (1995) Erwerb von Fuzzy-Wissen aus Selbstorganisierenden Neuronalen Netzen, Proc. 3. Workshop Fuzzy-Neuro-Systeme'95, Darmstadt/F.R.G, Nov. 15–17, 1995, 325–332.Google Scholar
  33. [33]
    Ultsch, A., Korus, D. (1995) Integration of Neural Networks with Knowledge-Based Systems, Proc. IEEE Int. Conf. on Neural Networks, IEEE-ICNN'95, Perth/Australia, Nov./Dec. 1995, Vol. 4, 1828–1838.Google Scholar
  34. [34]
    Ultsch, A., Mantyk, R., Halmans, G. (1993) A Connectionist Knowledge Acquisition Tool CONKAT, in D. J. Hand (Ed.) Artificial Intelligence Frontiers in Statistics, Chapman & Hall 1993, 256–263.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1997

Authors and Affiliations

  • A. Ultsch
    • 1
  • T. O. Kleine
    • 2
  • D. Korus
    • 1
  • S. Farsch
    • 1
  • G. Guimarães
    • 1
  • W. Pietzuch
    • 2
  • J. Simon
    • 2
  1. 1.Neuroinformatics and Artificial Intelligence, Dep. of MathematicsUniversity of MarburgMarburgGermany
  2. 2.Dep. of NeurochemistryMedical Centre of Nervous Diseases, University of MarburgMarburgGermany

Personalised recommendations