Abstract
Two types of decision rule for metrical diagnosis are presented in this paper. One of these is a method of simplified linear decision functions, which permits reliable results when suitable variables corresponding to the symptoms are chosen. The other type of decision rule is a method of self-organization of decision functions, which is a procedure to obtain reliable decision functions.
As an application of these methods, experiments of metrical diagnosis for congenital heart diseases are conducted. Further, as the hardware of decision procedure, an optical computer for linear decision functions and an electric computer for logical decision functions are proposed. In the conclusion of the paper, the problems of errors in decision processes are discussed.
