The paper introduces the Prevention Assistant, a software system that provides support for prevention activities in the setting of the patient-physician relationship. The purpose is to close around the patient an additional control loop able to supply personalized prevention messages based on the sensed health state and health determinants. The factors important for effective prevention initiatives are analyzed and an ontology is introduced to conceptually model them. Further, a mathematical model is used to put forward a solution for the fundamental concern of disease risk assessment in individuals about whom only few or old data is available. State-of-the-art statistical methods are employed in this sense, to estimate a probable value for unknown parameters.