Siegenthaler, Differential Diagnosis in Internal Medicine (ISBN), © Georg Thieme Verlag. Contents. 1−3. General Differential Diagnosis. 1. First published: September‐October forfindsebullperf.tk tbx. About. Related; Information. ePDF PDF. PDF · ePDF PDF · PDF. Febr. Krankheiten. W. Siegenthaler (Herausgeber): Differentialdiagnose innerer Krankheiten. Request Full-text Paper PDF. Citations (0).
|Language:||English, Spanish, Japanese|
|Genre:||Fiction & Literature|
|Distribution:||Free* [*Registration needed]|
Sept. PDF | Dieser Artikel stützt sich auf die Dissertation von Evers J. Zur Differentialdiagnose unklarer Fieberzustände, Köln, , und auf einen Fortbildungsvortrag von Gross R. auf der Chapter. Jan W Siegenthaler. Russi EW, Siegenthaler W, Lüthy R () Pneumocystis Differentialdiagnose: Innerer Krankheiten – vom Symptom zur Diagnose. SÄURE BASE STÖRUNGEN. Ein praktischer Approach zur Differenzialdiagnose. PD Dr. Thomas Fehr. Klinik für Nephrologie. UniversitätsSpital Zürich. Topics.
Table 3. The established inference engine searching valid derivations from given premises. The general is successful. So far, KONSDED has indicated about 20 inconsistencies in the medical knowledge base thinking errors and documentation mistakes and several hundred proposals for new relationships but only a few dozen of those have been used; the remaining longer part of proposals turned out to be redundant, e.
For generating diagnostic hypotheses, the concept of precalculating unique symptom patterns already explained in Section 2 is applied in an extended form in CADIAG Schwarz in  did an extended investigation of the usefulness of unique symptom patterns and their applicability generating diagnostic hypotheses. Schwarz found that between 40 and ; of the calculated symptom patterns were unique for one disease, i.
This great number of unique symptom patterns is very suitable for generating hypotheses. Reasons for that are: It could also be shown that the larger the number of documented diseases, the smaller the number of unique symptom patterns. Table 2 shows a segment of the documented symptoms and their relationships to a rheumatic disease and Table 3 a unique symptom pattern for that disease.
Given a certain symptom pattern, confirmed and excluded diagnoses, diagnostic hypotheses, and possible diagnoses are established. Confirmed diagnoses are obtained if one of the following conditions is true: Excluded diagnoses, on the other hand, are received to diseases; by the following: Diagnostic hypotheses are calculated by means of unique symptom patterns matching symptoms observed on the patient.
Possible diagnoses are made on the basis of preferential symptoms exhibited by the patient and selected as such by the diagnostician. The concept of preferential symptoms gives the physician the opportunity to propagate the diagnostic process in different directions and thus to broaden the diagnostic field. Preferential symptoms generate all diagnoses as possible diagnoses to which they possess FN-relationships.
It is advisable to select only symptoms as preferential symptoms which seem to have a certain importance-mostly those showing low degrees of ambiguity.
Unexplained symptoms of the patient under consideration are symptoms having relationships to neither confirmed diagnoses, diagnostic hypotheses nor possible diagnoses.
The repetition of the diagnostic process with unexplained symptoms offers a second possibility to explain every symptom of the patient. Extended explanations of the diagnostic results are given to the physician. This makes the diagnostic process comprehensible and supports trust in the computer-generated outcome. They allow an iterative diagnostic process and enable the physician to confirm or exclude diagnoses step by step.
Thus, it is precisely advised which examinations to perform next. This fact can be seen as an educational tool to optimize the examinations necessary and sufficient to perform. Four hundred and twenty-six cases from a rheumatological hospital were tested. About symptoms, signs, test results, and findings among them about present and about absent were available for each case. The results of the cases are shown in Table 5. Forty-seven cases with pancreatic diseases from a university clinic were tested.
About symptoms, signs, test results, and findings among them about 30 present and absent were available for each case. The results are shown in Table 6. By doing this, one avoids problems having their cause in: But, on the other hand, a clear distinction between soft relationships can very often be found. Naturally, a medical expert values these relationships differently when making his diagnosis.
Starting from this consideration a useful tool to formalize soft expressions was found in the theory of fuzzy sets. Fuzzy set theory developed by Zadeh in  see also [49,50] with its ability of defining inexact medical entities as fuzzy sets, with its linguistic approach [Sl] providing an excellent approximation to medical texts as well as its power of approximate reasoning [52, seems to be perfectly appropriate for designing and developing medical expert systems.
Reviews of fuzzy approaches to medical decision making are given in  and . An early attempt to computer-assisted medical diagnosis using fuzzy set theory that can be considered as a preliminary version of CADIAG-2 was published in . Frequency of occurrence and strength of confirmation are considered to be linguistic variables see Bellman and Zadeh . These linguistic variables can take the following linguistic values: Structure of CADIAG-2 with connection to a medical information system dashed lines mark components effective before starting the individual consultation.
Tusch [9, lo] uses this model in a slightly modified form for the cranial computer tomography. The application considers five tumor diagnoses: Twenty-five symptoms gathered by seven different examinations describe each case: Eight hundred and two tumor diagnoses were used to perform this calculation.
Introduction In the final version of CADIAG-2, the compositional rule of inference proposed by Zadeh  and introduced in medical diagnosis by Sanchez [SS, has been selected to calculate the membership grades of patients to diseases. The relationships between symptoms and diseases are described by occurrence and confirmation values of either linguistic, statistical, or judgmental origin. Furthermore, complex combinations of symptoms that can be evaluated by means of fuzzy logical connectives show relationships to diseases.
They take their values ps, in [0, l] u -. The values ps, indicate the degrees of membership of symptoms Si to patients P,. The essential advantage of this formal approach is the possible representation of borderline symptom values. A detailed interpretation of symptom fuzzy values pLs,is shown in Table 7. Interpretation of symptom medical fuzzy values ps, Fuzzy values flc Interpretation 0.
S, lies between range. Interpretation present at P,. Th ere are criteria that exclude D, as diagnosis. Dj is regarded as diagnostic hypothesis. P, shows IC, or fulfilled at P,.
IC, or SC, cannot be determined because of symptoms, diseases, or intermediate combinations not yet examined or determinable.
Table Diseases or diagnoses are treated in a similar way see Table 8. Intermediate and symptom combinations can also have fuzzy logical values see Table 9. They contain symptoms, diseases and, in case of symptom combinations, if necessary, intermediate combinations, as fuzzy logical variables. The appropriate fuzzy logical connectives are presented in Table Every single relationship is characterized by two aspects: Reasonable numerical representatives for I, and 1, were chosen to simplify fuzzy inferences.
Table 11 shows the linguistic terms and their numerical representatives to describe the frequency of occurrence and the strength of confirmation of one medical entity for another. Table 12, analogously to Table 2, shows a segment of documented symptoms and their frequency of occurrence and strength of confirmation to the rheumatic disease ankylosing spondylitis.
Segment Ankylosing of documented approaches to computer-assisted symptoms and spondylitis medical their relationships in CADIAG-2 diagnosis for the rheumatic disease ankylosing spondylitis Symptoms Spine. Spine, Spine, Spine, Spine. Spine, Spine. Spine, Spine, Spine. X-ray, X-ray. X-ray, spine, spondylitis spine, arthritis of sacroiliacal spine, bamboo-spine i 4cm cervical, restriction of motion thoracic, restriction of motion calcification of longitudinal ligament ankylosis.
Detailed checks for contradictions in the presented symptom pattern and the computed patterns of intermediate and symptom combinations are performed.
Afterwards, a differential diagnostic group can be chosen. Then, confirmed diagnoses are determined. The criteria for obtaining confirmed diagnoses are as follows: Diagnostic hypotheses are obtained by considering the following criteria: OYJ 4 8O. OY; 4 Unexplained symptoms, detailed explanations of the diagnostic results, and proposals for further examination of the patient are indicated in a way similar to that of CADIAG An additional feature is built into the explanation procedure of diagnostic hypotheses.
Because the value pi,] calculated by a max-min composition is independent of the number of symptoms or symptom combinations that can be applied for Dj, a heuristic point number is counted that takes into account the number of symptoms or symptom combinations supporting the hypothesis.
The rheumatic and pancreatic cases described in section 3. The results are shown in Tables 14 and Reasons for failure in diagnosing rheumatic diseases are in general the same as mentioned in Section 3. Reasons for failures occurring in diagnostic results for pancreatic diseases are: In general, it can be claimed that CADL4G-2 is quite capable of handling those aspects which are not only strongly characterizing medical knowledge but also real world knowledge such as: Both systems are general medical expert systems.
They are directly connected with the hospital information system of the University of Vienna Medical School.
The tests are not yet finished because of the extended medical knowledge bases containing about diseases where about symptoms, signs, test results, and findings are considered. Special emphasis with establishing CADIAG-1 and CADIAG-2 was given to rare diseases, which the individual physician may not keep in consideration, as well as to detailed reasonings of proposed or excluded diagnoses, low cost plans for further investigations on the patient, and pathological signs not yet explained by the diagnostic results.
Acknowledgements-The authors gratefully thank W. Bogad, M. Hatvan, MSc. Grdger, G. Sedivy and F. Lipomersky for their participation in this project, in particular for their immense work in programming parts of the systems and handling the acquisition of medical knowledge and the documentation of patient data. Furthermore our thanks are given to G. Ginzler who, as the responsible expert for systems software at the Computer Center of the University of Vienna Medical School, has supported this project from the beginning.
Ledley and L. Lusted, Reasoning foundations 2. Spindelberaer and G. Erfahrungen, Projekte. Edited by K. Fellinger ,pp. Verlag Brider Hollinek, Wien Gang], G. Grabner and P.
Adlassnig and G. Grabner, Approaches to computer-assisted diagnosis in gastroenterology. Adlassnig, G. Kolarz, F. Lipomersky, I. Groger and G.
Barber, P. Reichertz and F. Roger , pp. Springer, Berlin Adlassnig, A fuzzy logical model of computer-assisted medical diagnosis, Meth. Gupta and E. Sanchez , pp. NorthHolland, Amsterdam Brauer , pp. Springer, Berlin I. Its principles and values, Grabner, The Viennese computer-assisted Automedica 3, Medizinisches Diagnosesystem unter Verwendung von Fuzzy Adlassnig, Ein Computerunterstiitztes Teilmengen, Ph. Pitman Books, London Barr and E.
Editors , The Handbook of Artificial Intelligence. Pitman Books, Cohen and E. Feigenbaum London Nilsson, Principles of Artificial Intelligence. Tioga, Palo Alto Amarel and A. Safir, A model-based method for computer-aided medical Weiss, C. II, Kulikowski and A. Safir, Glaucoma consultation by computer. M YCIN. Elsevier, New York Shortliffe, Computer-Based Medical Consultation.
Yu, L. Fagan, S. Wraith, U.
Clancey, A. Scott, J. Hannigan, R. Blum, B. Buchanan and S. Cohen, Antimicrobial selection by a computer-a blinded evaluation by infectious disease experts. Pople, J. Myers and R. Pople, The formation of composite hypotheses in diagnostic problem solving-an exercise in synthetic reasoning, in Proceedings 01 the 5th International Joint Conference on Artificial Intelligence.
Miller, H. Szolovits and S. Weiss, K.
Kern, C. Kulikowski and M. Weiss and C. Stanford University. Knowledge engineering for medical decision making: Sort by: Influence Recency. Highly Influenced. Powers ArXiv Wang Innovations in Systems and Software Engineering Where Do We Stand in ? Gupta , Qamar Rehman , Mridula Verma Formal approaches to rule-based systems in medicine: Reasoning Citation Statistics Citations 0 10 20 '86 '93 '01 '09 ' Semantic Scholar estimates that this publication has citations based on the available data.
See our FAQ for additional information. References Publications referenced by this paper. Barr and E. Feigenbaum Editors , The Handbook…. Linguistic variables, approximate reasoning and dispositions.
Lotfi A. Ammann, Ikterus, in Differentialdiagnose innerer….