Abstract
Major depressive disorder has been modeled as a complex dynamical system. In this study, daily emotional changes in personalized depression were investigated using the concepts of fusion of recurrences and tensor decomposition. This study aims to answer the question if the use of the proposed approach can objectively discover what state of mind is considered as the most important attribute for differentiating the effect of medical alteration in major depression. By applying two different fusion models for combining time series of negative and positive moods obtained from a unique dataset of a case of personalized clinical depression, and the decomposition of these fusions by means of a multiway data analysis, the results suggest that while the interfusion of both negative and positive moods can be complementary, the intrafusion of negative moods would maximize the differenfiation.