This project created virtual patients who can respond to hypothetical therapeutic interventions in an agent-based model of dementia management. We evaluated the overall response of patients by collecting statistics and observing their group behaviour. In this model virtual patients were actively seeking treatment for symptoms of depression associated with dementia. Responses to hypothetical therapeutic interventions consisted of both generic (common to all patients) and individual (modified for each patient) components. The preliminary results show that even simple sets of rules governing behaviour of virtual patients can lead to quite complex responses at the group level. Furthermore, the lessons learned from monitoring the group behaviour provided valuable feedback which is now being used to modify the creation of individual virtual patients e.g. implementation of histories of previous successful and unsuccessful treatments.