A Comparison of the Measure of Surprise Between Several Variables for Medical Control

Abstract

The fuzzy causation measure K has be defined using the fuzzy Subsethood theorem. It is a measure of the role of unknown factors in the determination of a change in cardinality of a fuzzy set. Methods: We measured the value of K for: (1) change in fuzzy set cardinality using low density and high density lipoprotein values in 10 patients with history of ischemic stroke, (2) the change in fuzzy set cardinality for expert opinion regarding the degree of goal value attained by same variables, and (3) the change in fuzzy cardinality for non-expert grading of degree of control of the same variables. We compared K values for change in lab results, expert results and non expert results. Results: The degree of change in K for low and high density lipoprotein values in each of 10 patients, and the non expert was minimal compared to that of the expert's opinion. Conclusion and Interpretation: The expert and the non expert use their own normalization values for determination of degree of clinical goal values met for a given laboratory result. In the case of the expert, this normalization changes based on his experience and in a non linear fashion. For the non expert who normalizes according to a set standard of written clinical guidelines, there is no change in K reflecting a rigid standard canceling out the effect of any other contributing factors on his judgment.