Medical expert system on fuzzy analog ganglionar lattices: explication and prospection based on sensitivity

A reasoning method to explicate the conclusions-a posteriori- and to prospect for the next test-a priori-applied to medical expert systems based on fuzzy analog ganglionar lattices is presented. This method is founded on the sensitivity criterion of the consequent respect to the antecedents. It proves to be suited for problems of any complexity, specially when they are formulated on well-established intermediate concepts, such as syndromes, clinical finding, laboratory procedures, etc., as is usually the case in medical procedures. This type of expert system uses the structural properties of the ganglionar lattice to produce explications for its conclusions at different levels of abstraction, and to an arbitrary, but fixed, explicative degree. A measure to evaluate the consequent's achieved preciseness (certainty) is also supplied. A full example of application in the diagnosis of cardiac insufficiency with hypertensive myocardiopathy is shown, and a computational implementation of this analog procedure is described. Finally, the reasoning methods for explicating and prospecting are discussed.