Shimizu H, Haida M, Niwa M
Department of Medical Informatics, School of Medicine, Tokai University, Kanagawa Japan.
Tokai J Exp Clin Med. 1990 Mar;15(1):51-79.
Based on previous studies, a consultation program using clinical laboratory data has been constructed. After converting all data into the standard deviation index (SDI), the score for preregistered diseases was calculated by integrating products of SDI and the weighting factors (WF) pre-determined empirically by medical experts for each item. Ten diseases in order of their maximum scores, the most predictable ones, and additional comments were printed out. The principle of our consultation program is based on the modified Bayesian theory which employs, instead of probability a posteriori, weighting factors (WF) corresponding to the grade of membership in the "fuzzy set" concept. Evaluations based on this interpretation performed 2-3 times during the clinical course in nine patients were found to agree qualitatively with each clinical course and diagnosis. This report also contains 1) an outline of the progress of this consultation program in Japan and abroad, 2) the actual level of acceptance of this system by physicians, 3) and the characteristics and problems of our program. We are now planning to report a quantitative evaluation of this program using many cases in some specified field, instead of one case of each disease.
