Multi-Agent Simulation of Diagnostic and Logistic Processes in Hospitals

Hospitals can be viewed as complex systems with sophisticated and often hardly control-able dynamics. Until recently the need for effective management was not really urgent, as the health insurances were able to fund all the necessary tasks in a sufficient manner. But the pressure grows that hospitals have to organize all processes in an optimal way to save time and thus expenses. In medical environments, like hospitals, this afford is twofold. On one side this refers to diagnostic processes targeting on a secure determination of what disease the patient is suffering from and the management of a successful therapy. On the other side time and resource management techniques are necessary for an efficient schedule of care and examination tasks. Because of the unpredictability and dynamics in medical domains the logistic processes can hardly be controlled automatically using hospital management systems. Thus there is a high potential of possible improvement especially for logistic processes that is not accessible with traditional techniques that are used in e. g. production systems. Using multi-agent simulation the dynamics in a hospital can be mapped into a model containing more or less sophisticated agents with arbitrary complex behavior. Different events have not to be modeled using probability distributions that are based on observations from a global point of view, but the complete " life span " of a patient with all tasks from diagnosis via therapy to the discharge can be modeled explicitly and simulated in detail. In this scenario multi-agent simulation provides a flexibility that neither petri nets nor process-based simulation are capable of. Thus the dynamics in an hospital can be better formulated and comprehended leading to the possibility of repeated " what-if " simulations with questions and answers to check the influence of specific factors. It might become possible to generate semi-automatically suggestions for optimizing certain parameters, for example the personal equipment regarding distinct criteria like minimal overtime. A multi-agent simulation model of In: Tagungsband zum Workshop "Intelligente Softwareagenten und betriebswirtschaftliche