The relationship between doctors and their patients is gaining more and more importance in the health care providing. It determines the compliance of the treatment and a part of the curative process. In the psychiatry the therapeutic relationship has even more power. Therefore having a general rule that could guide doctors towards a good relation with their patients would be very useful. This paper describes experiments in automated acquisition of such a rule by means of data mining techniques. Decision rules are provided to analyze the patients' satisfaction regarding the therapeutic treatment and to recognize whether a patient belongs to a psychiatric ward or not. Data mining techniques show very satisfactory results on the experiments done with four different algorithms.