Multi-Agent Architecture for Medical Diagnostic Systems

Article is devoted to the development of a multiagent architecture of a medical system for automated diagnosis of human health. Due to the ability to create transparent and clear models, the multi-agent approach allowed us to develop an easily scalable system that implements a set of scenarios for assessing the functional state of the human body, depending on the purpose of medical research and also available indicators of the human body state. The proposed multi-agent architecture of the information system for automated diagnostics is universal. On the one hand, it allows you to implement the standard functions of collecting, processing and storing information. On the other hand, special diagnostic functions are implemented based on a unique technique developed by the authors. The uniqueness lies in the proposed technology for integrating multi-format information based on cluster and discriminant analysis, as well as neural network models. The paper considers the modules of the system, each of which is a multi-agent system, represented by a combination of agents and production rules that describe the interaction of agents.