Identifying decision maker's preferences through a prototype based inductive learning method: A medical case study

We present a model of a prototype based inductive method supporting the modelling and identification of decision maker's preferences. We approach decision making using concept formation techniques, namely the prototype model, and identify decision making preference via a two-phase process. During the first phase we identify crisp decision patterns while in the second we focus on patterns that are influenced by decision-making context. Descriptions that associate both types of decision patterns are induced using machine learning. We demonstrate our model using a medical domain that relates to therapeutic planning of maldescensus testis in children. Finally we explore advantages and limitations of our method and discuss the potential for further research on the subject.