Agent driven diagnosis in medicine

Abstract

Embedding Machine Learning technology into Agent Driven Diagnosis Systems adds a new potential to the realm of Medicine, and in particular to the imagiology one. However, despite all the research done in the last years on the development of methodologies for designing MultiAgent Systems (MAS), there is no methodology suitable for the specification and design of MAS in complex domains where both the agent view and the organizational view can be modelled. Current multi-agent approaches either take a centralist, static approach to organizational design or take an emergent view in which agent interactions are not pre-determined, thus making it impossible to make any predictions on the behavior of the whole systems. Most of them also lack a model of the norms in the environment that should rule the behavior of the agent society as a whole and/or the actions of individuals. In this paper, we propose a framework for modelling agent organizations, and we illustrate the different components of a society with one modality, the Axial Computed Tomography scenario, combining two methodologies for problem solving, the Artificial Neural Networks and the Case Based Reasoning ones.