The SIMON Architecture: Distributing Data, Tasks, and Knowledge in Intelligent ICU Monitoring Systems

Otherwise successful research efforts in intelligent critical care monitoring often fail to achieve the same level of success when applied to actual clinical systems, or when additional tasks or knowledge, possibly outside the original problem domain, are added. A variety of factors contribute to this lack of success, including issues related to system reliability, scalability, and inherent design restrictions imposed by the underlying knowledge model of the system. These issues can and, in part, have been addressed by distributing data and monitoring tasks to separate modules. However, effective knowledge distribution remains a significant challenge. This paper reviews research efforts to distribute intelligent monitoring systems, and describes the SIMON (Signal Interpretation and MONitoring) architecture, which currently supports distributed tasks and data. Advantages and disadvantages of this distribution are detailed, and preliminary results of clinical testing are presented.