Gernot Kolarz

CADIAG: Approaches to computer-assisted medical diagnosis

CADIAG-1 is a medical expert system, based on a symbolic logic representation of medical relationships. Strong relationships such as confirming, excluding or obligatory occurrence are applied to confirm or exclude diagnoses. Weak relationships are represented by facultative and not confirming relationships (FN-relationships). Diagnostic hypotheses are established by systematic combination of symptoms showing FN-relationships. CADIAG-2, a medical expert system based on fuzzy set theory and fuzzy logic, allows detailed specification of medical relationships. Here the diagnostic process also provides confirmed and excluded diagnoses as well as diagnostic hypotheses. Hypotheses are calculated by considering fuzzy relationships between medical entities. 426 cases with rheumatic and 47 cases with pancreatic diseases were tested. For CADIAG-1, the overall accuracy for confirmation and hypothesis generation is calculated with 91.1% for rheumatic diseases and 100% for pancreatic diseases. CADIAG-2 reached an overall accuracy of 93.7% for rheumatic cases and 91.5% for pancreatic cases.

Representation and semiautomatic acquisition of medical knowledge in CADIAG-1 and CADIAG-2

CADIAG-1 and CADIAG-2 (Computer-Assisted DIAGnosis) are medical expert systems especially designed for ill-defined areas such as internal medicine. Both systems are being tested in the setting of a medical information system. With respect to their knowledge representation, CADIAG-1 has obvious advantages in totally ill-defined areas such as syndromes in internal medicine, whereas CADIAG-2 seems more suited for domains with basic laboratory programs, e.g., hepatology or gall bladder and bile duct diseases. The formalization of relationships between medical entities led to first-order predicate calculus formulas in the case of CADIAG-1 and to a model based on fuzzy set theory in the case of CADIAG-2. In both systems two kinds of relationships between medical entities are considered: (1) necessity of occurrence and (2) sufficiency of occurrence. Statistical interpretations using the 2 X 2 table paradigm yield a way to calculate these relationships automatically from samples of patient data. Results obtained by exploiting 3530 patient records from a rheumatological hospital are presented. The described application is a machine-learning program that allows inductive learning from examples under statistical uncertainty.

Problems in establishing the medical expert systems CADIAG-1 and CADIAG-2 in rheumatology

CADIAG-1 and CADIAG-2 are medical expert systems with applications in rheumatology, gastroenterology, and hepatology. CADIAG-1 is based on a symbolic logic representation of medical relationships between symptoms, signs, or findings and diseases. Definite relationships (obligatory occurrence, confirming, and excluding) as well as uncertain relationships (facultative occurrence and not confirming) are applied to confirm or exclude diagnoses and to establish diagnostic hypotheses. CADIAG-2 employs fuzzy set theory and fuzzy logic to formalize medical entities and relationships. The medical concept of confirming or excluding diagnoses is identical to that of CADIAG-1, but diagnostic hypotheses are generated differently. Here, a documentation of medical relationships allowing gradual transitions from “always” to “never” for the frequencies of occurrence of symptoms with and from “strong” to “weak” for their strengths of confirmation for diseases leads to strongly or weakly supported diagnostic hypotheses in the actual case. Tests with 322 real patient cases from a rheumatological hospital, each including between 500 and 700 symptoms, signs, and findings, were carried out. The percentage of cases diagnosed correctly is about 80%. Problems and pitfalls that became apparent in the evaluation of the cases are shown and discussed.

Evaluation of two different models of semi-automatic knowledge acquisition for the medical consultant system CADIAG-II/RHEUMA

As part of a plan to promote semi-automatic knowledge acquisition for the medical consultant system CADIAG-II/RHEUMA, this study sought to explore and cope with the variability of results that may be anticipated when performing knowledge acquisition with patient data from different patient settings. Patient data were drawn both from a published study for the classification of rheumatoid arthritis (RA) and from a large database of rheumatological patient charts developed for the CADIAG-II/RHEUMA system. An analysis of the relationships between RA and selected CADIAG-II/RHEUMA symptoms was done using two models. In one of them, we controlled for the differences in baseline frequencies of symptoms and diseases in the two study populations as an important factor influencing the results of the calculations. Other factors that were identified included inconsistent definitions of symptoms and diseases, and the different composition of study groups in the two study populations. By eliminating differences in baseline frequencies as the most important bias, the results obtained from the two different knowledge sources became more consistent. All remaining inconsistencies and uncertainties about the contribution and relative importance of the factors were formalized using fuzzy intervals.

On the applicability of diagnostic criteria for the diagnosis of rheumatoid arthritis in an expert system

Abstract Cadiag -2/ Rheuma is a medical expert system developed to assist in the differential diagnosis of rheumatic diseases. Based on fuzzy set theory and fuzzy logic, it supports the formalization of vague and uncertain medical information (i.e., medical entities and relationships between them) and draws justifiable conclusions from these imprecise data. Given a patients finding patter, Cadiag -2 provides confirmed and excluded diagnoses, diagnostic hypotheses, and suggestions for further examinations. The knowledge base of Cadiag -2 has been designed to contain simple finding/disease relationships as well as diagnostic rules of high complexity to confirm or hypothesize disease. We shall present results obtained with 300 clinical cases from a hospital for rheumatic diseases. Different rules for the diagnosis of rheumatoid arthritis based upon classification criteria issued by the American Rheumatism Association were tested against each other. That diagnostic rule which had shown the best results was then further improved by a rheumatology expert, which finally yielded a sensitivity of 83.3% and a specificity of 95.3%.

Knowledge Acquisition Study and Accuracy Rate Evaluation for CADIAG-2/Rheuma With 308 Clinical Cases

Cadiag-2/Rheuma is a medical expert system assisting in the differential diagnosis of rheumatic diseases. The aim of this study was to establish a set of criteria for diagnosing definite rheumatoid arthritis (RA) that provides optimal accuracy and to implement this set of criteria as an IF- THEN rule for application in Cadiag-2/RHEUMA. First, two different sets of criteria for the classification of definite RA described in medical literature were implemented and their respective diagnostic accuracies were evaluated with 154 patients suffering from RA and 154 control subjects. Second, that set of criteria which had performed best became the starting point for establishing an improved set of diagnostic criteria that eventually reached an accuracy of 88.7% (81.8% sensitivity and 95.5% specificity). This improvement was possible by combining literature definition of RA with specific clinical experience of a rheumatology expert.