The main objective of the project is to enhance the current generation of Intelligent DEcision Support systems (INDECS) through the use of multivariate statistical analysis. It includes a study of the structure of relationships between components of multidimensional random vectors; the development of a theory of T-normal distributions; the construction of models for the description of sequential responses of random objects; the relation between multivariate statistical analysis and Bayesian belief networks; and a study of the stability of multivariate statistical methods. Another objective of the project is to investigate the applicability of the techniques developed to two different problems: the prediction of medical treatment of children in states of emergency, and the classification of ecological states of arable land.