Multicriteria Decision Aid For Helping To Understand Health Care Consumption

Two years ago, the Belgian Ministry for Social Affairs asked the departments of Operational Research (SMG) and Applied Economics (DULBEA) of the Université Libre de Bruxelles to develop methodological tools aimed at helping to understand health care budgets consumption. The first step of this research concerns the identification of “pathological channels” defined from mathematical programs and through the correlation between cares expenditures time series. The paper will deal with the second step, which is dedicated to the construction of a multicriteria fuzzy assignment procedure of the insured to these channels. The multicriteria sorting problematic, also called P.β problematic (in contrast to the P.α choice problematic and the P.γ ranking problematic), consists in assigning a set of alternatives, described by their profile of performances according to k criteria, to some predefined categories characterised by one or several reference point(s), also called prototype(s). The assignment to a category is based on the type of relation existing between the action to be assigned and the category prototype(s). When applying these concepts to the analysis of Belgian health care budgets, the concordance and non-discordance aggregation functions of the available methods and software (mainly inspired by the crisp or valued outranking approach, and more precisely by the ELECTRE methods) did not seem appropriate to solve the problem. Therefore, inspired by the particular characteristics of our application, we investigate the definition of multicriteria weighted operators aggregating a valued non compensatory (conjunctive) measure of similarity in the case of non ordered categories for which the criteria rather look like assignment constraints.