FIELD: medical technology.
SUBSTANCE: invention relates to a means for recognizing the operating state of a sensor for continuous monitoring of glycemia. The process starts by receiving continuous monitoring data related to sensor performance and containing compressed monitoring data. A trained learning algorithm is provided to recognize the operating state of the sensor, which characterizes the sensor function, and the learning algorithm is trained in accordance with a training data set containing historical data and compressed training data. Then the operating state of the sensor is recognized by analyzing continuous monitoring data using a trained learning algorithm. The output data is provided indicating the recognized operating state of the sensor. Historical data consists of data collected, recorded and / or measured prior to the process of recognizing the operating state. Compressed monitoring data and compressed training data are obtained by linear regression and / or smoothing as a result of reducing their size, and at different stages of compression, the monitoring data and training data contain data per second, data per minute and / or statistical data, including characteristic values such as sensor parameters, variance, noise, or rate of change.
EFFECT: technical result is to increase the likelihood of detecting a potential problem with the operating state of the sensor.
12 cl, 9 tbl, 14 dwg
