This paper is involved with the analysis of serial measures concerning a set of patients. For every patient, a variable number of series is measured, in different periods along time. A usual way to analyze this kind of information is to reduce all series belonging to the same patient to some synthesis measure, as a mean curve per patient. However, this kind of synthesis measures involve some implicit hypothesis that often are not taken into account. If those hypothesis don’t hold, the results of analyzing the synthesis curves may be very far from reality. In this paper, a real data set concerning psychofisiological effects of electroconvulsive therapy which includes repeated serial measures per patient is analysed in two ways: first, using the classical approach of analysing one mean curve per patient; secondly, using a new methodology called KDSM. After all, results obtained in both cases are compared and some conclusions raise.
Key words: Knowledge discovery, Serial measures, Clustering and Electroconvulsive therapy