Computer-Aided Diagnosis in Jaundice: Comparison of Knowledge-based and Probabilistic Approaches

Abstract

The study reported in this paper is aimed at evaluating the effectiveness of a knowledge-based expert system (ICTERUS) in diagnosing jaundiced patients, compared with a statistical system based on probabilistic concepts (TRIAL). The performances of both systems have been evaluated using the same set of data in the same number of patients. Both systems are spin-off products of the European project Euricterus, an EC-COMACBME Project designed to document the occurrence and diagnostic value of clinical findings in the clinical presentation of jaundice in Europe, and have been developed as decision-making tools for the identification of the cause of jaundice based only on clinical information and routine investigations. Two groups of jaundiced patients were studied, including 500 (retrospective sample) and 100 (prospective sample) subjects, respectively. All patients were independently submitted to both decision-support tools. The input of both systems was the data set agreed within the Euricterus Project. The performances of both systems were evaluated with respect to the reference diagnoses provided by experts on the basis of the full clinical documentation. Results indicate that both systems are clinically reliable, although the diagnostic prediction provided by the knowledge-based approach is slightly better.

Keywords

Expert Systems - Evaluation - Jaundice Diagnosis - Medical Decision Making

• REFERENCES

• 1 Van Gennip EMSJ, Talmon JL. eds Assessment and Evaluation of Information Technologies in Medicine. Amsterdam: IOS Press; 1995
• 2 Van der Loo RP. Overview of published assessment and evaluation studies. In: Van Gennip EMSJ, Talmon JL. eds Assessment and Evaluation of Information Technologies in Medicine. Amsterdam: IOS Press; 1995: 261-81.
• 3 Talmon JL, Van der Loo R. Literature on assessment of information technology and medical KBS evaluation. studies and methodologies. In: Van Gennip EMSJ, Talmon JL. eds Assessment and Evaluation of Information Technologies in Medicine. Amsterdam: IOS Press; 1995: 283-327.
• 4 Martin WB, Apostolakos PC, Roazen H. Clinical versus actuarial prediction in the differential diagnosis of jaundice. Am J Med Sci 1960; 240: 571-8.
• 5 Burbank F. A computer diagnostic system for the diagnosis of prolonged undifferentiating liver disease. Am J Med 1969; 46: 401-15.
• 6 Cattaneo AD, Lucchelli PE, Rocca E, Mattioli F, Becchi G. Computer versus clinical diagnosis of biliary tract diseases. J Abd Surgery 1972; 14: 71-5.
• 7 Knill-Jones RP, Stern RB, Girmes DH, Maxwell JD, Thompson RPH, Williams R. Use of sequential Bayesian model in diagnosis of jaundice by computer. Br Med J 1973; i: 530-3.
• 8 Stern RB, Knill-Jones RP, Williams R. Use of computer program for diagnosing jaundice in district hospitals and specialized liver unit. Br Med J 1975; ii: 659-62.
• 9 Wheeler PG, Theodossi A, Pickford R, Laws J, Knill-Jones RP, Williams R. Non-invasive techniques in the diagnosis of jaundice ultrasound and computer. Gut 1979; 20: 196-9.
• 10 Lindberg G. Studies on Diagnostic Decision Making in Jaundice. (Thesis). Stockholm: Karolinska Institute; 1982: 1-60.
• 11 Matzen P, Malchow-Møller A, Hilden J. et al Differential diagnosis of jaundice; a pocket diagnostic chart. Liver 1984; 4: 360-71.
• 12 Saint-MarcGirardin M-F, Le Minor M, Alperovitch A, Roudot-Thoraval F, Metreau J-M, Dhumeaux D. Computer-aided selection of diagnostic tests in jaundiced patients. Gut 1985; 26: 961-7.
• 13 O’Beirn SF, Given FH. A general hospital diagnostic database on jaundice. In: Tsiftisis D. ed Objective Medical Decision-making: Systems Approach in Disease. Workshop Proc; 1985: 103-7.
• 14 Boom R, Gonzalez C, Fridman L. et al Looking for “indicants” in the differential diagnosis of jaundice. Med Decis Making 1986; 6: 36-41.
• 15 Malchow-Møller A, Thomsen C, Matzen P. et al Computer diagnosis in jaundice. Bayes’ rules founded on 1002 consecutive cases. J Hepatol 1986; 3: 156-63.
• 16 O’Beirn SF, Given HF, Lavelle SM. Experiences with compiling a diagnostic database on jaundice presenting to a general hospital. Irish J Med Sci 1987; 156: 93-7.
• 17 Lindberg G, Thomsen C, Malchow-Møller A, Matzen P, Hilden J. Differential diagnosis of jaundice: applicability of the Copenhagen Pocket Chart proved in Stockholm patients. Liver 1987; 7: 43-9.
• 18 Börsh G, Baier J, Glocke M, Nathasius W, Gerhardt W. Graphical analysis of laboratory data in the differential diagnosis of cholestasis: a computer-assisted prospective study. J Clin Chem Clin Biochem 1988; 26: 509-19.
• 19 Khoo US, Lavelle SM. The utility of ancillary tests in the diagnosis of jaundice. Med Inf 1988; 13: 93-104.
• 20 Malchow-Møller A, Thomsen C, Hilden J, Matzen P, Mindelholm L, Juhl E. A decision tree for early differentiation between obstructive jaundice and non-obstructive jaundice. Scand J Gastroenterol 1988; 23: 391-401.
• 21 Segaar RW, Wilson JHP, Habbema JDF, Malchow-Møller A, Hilden J, van der Maas PJ. Transferring a diagnostic decision aid for jaundice. Neth J Med 1988; 33: 5-15.
• 22 Fieschi M. Le système Sphinx. In: Fieshi M. ed Intelligence Artificielle en Médecine. Paris: Ed Inserm, Ed. Masson; 1986: 75-137.
• 23 Cammà C, Garofalo G, Almasio P. et al A performance evaluation of the expert system “Jaundice” in comparison with that of three hepatologists. J Hepatol 1991; 13: 279-85.
• 24 Euricterus Project Management Group. Objective medical decision-making: clinical database for diagnosis of jaundice (EURICTERUS). In: Beneken JEW, Thevenin V. eds Advances in Biomedical Engineering. Amsterdam: IOS Press; 1993: 35-64.
• 25 Molino G, Ballarè M, Aurucci PE, Ripa di Meana V. Application of artificial intelligence techniques to a well defined clinical problem: jaundice diagnosis. Int J Biomed Comp 1990; 12: 189-202.
• 26 Molino G, Molino F, Torasso P, Console L, Ardissono L, Pogliano P. Computer-assisted development of diagnostic expert systems. A domain-independent package (EMPTY) for acquisition and use of expert’s medical knowledge. Proc 16th SCAMC. Baltimore: 1992: 344-8.
• 27 Console L, Molino G, Ripa di Meana V, Torasso P. LIED Liver: Information, Education and Diagnosis. Meth Inform Med 1992; 31: 284-97.
• 28 Molino F, Molino G. Revisione e verifica preliminare di ICTERUS: un sistema esperto per la diagnosi clinica dei pazienti itterici. In: Serio A, Maceratini R. eds Proc 8th Nat Congr Med Inform, AIIM ‘94. Roma: 1st Ital Med Sociale; 1994: 199-205.
• 29 Molino G, Molino F. Assessment and evaluation of knowledge-based expert systems for medical diagnosis. In: Van Gennip EMSJ, Talmon JL. eds Assessment and Evaluation of Information. Amsterdam: IOS Press; 1994: 151-66.
• 30 Hilden J, Habbema JDF, Bjerregaard B. The measurement of performance in probabilistic diagnosis. II. Trustworthiness of the exact values of the diagnostic probabilities. Meth Inform Med 1978; 17: 227-37.
• 31 Hilden J, Habbema JDF, Bjerregaard B. The measurement of performance in probabilistic diagnosis. III. Methods based on continuous functions of the diagnostic probabilities. Meth Inform Med 1978; 17: 238-46.
• 32 Hanley JA, McNeil BJ. The meaning and use of the area under a Receiver Operating Characteristic (ROC) curve. Radiology 1982; 143: 26-36.