A note on the usefulness of linguistic variables for differentiating between some respiratory diseases (Q1071463)

An uncommon application of linguistic variables as predictor variables for obtaining classical discriminant functions is presented. Commonly only quantitative features are used as predictor variables for evaluating linear, quadratic or canonical discriminant functions because of the assumption of normality. This paper deals with results of discrimination between various respiratory diseases using quantitative and linguistic variables together. Linguistic variables in medical applications were obtained by \textit{L. Saitta} and \textit{P. Torasso} [ibid. 5, 245-258 (1981; Zbl 0455.92003)] as a conversion of the groups of original, discrete features. The presented method enables the simultaneous use of quantitative and qualitative features to the description of the patient's state. For this reason linguistic variables seem to be an efficient tool in medical applications.