In this paper we present a preliminary work applying machine learning techniques to the assessment of quality of life (QOL). In 2008 the Government of Catalonia introduced the GENCAT scale, a QOL questionnaire for dependent people users of the social and human services of Catalonia. Using data from a QOL research of 50 people with intellectual disabilities and/or mental illness of the Taller Jeroni de Moragas, we have applied a lazy learning method to discover relations between the different dimensions considered in the GENCAT scale. Our goal is to provide a basis to refine the model of QOL in a way that could support general intervention programs and a better understanding of the necessities of dependent people. This study is an interdisciplinary research of computer scientists, psychologists and human service practitioners.