Architecture and principles of routing patient monitoring data in the medical internet of things system based on machine learning

IoT devices in telemedicine systems generate huge data flows, while the message intensity is a non-stationary flow characterized by periods of peak loads and periods of low message intensity. Processing such flows, even taking into account the principles of buffering and redistribution of flows, causes delays in the provision of telemedicine services to end users in the IoT cloud environment, which is unacceptable for critical medical applications. The article discusses the features and limitations of existing systems for collecting and pre-processing telemedicine data in the context of an increasing number of subscribers. It is proposed to use fuzzy clustering procedures for message packet identifiers operating at the level of intermediate calculations of Fog nodes, thereby reducing delays from minutes to milliseconds.