Most existing physiological testing systems broadly classify monitored physiological data into three categories: normal, abnormal, and highly abnormal, but do not consider differences in the importance of data within the same category, which may result in the loss of data of higher importance. In addition, the purpose of physiological monitoring is to detect health abnormalities in patients earlier and faster, thus enabling risk avoidance and real-time rescue. Therefore, we designed a system called the adaptive physiological monitoring and rescue system (APMRS) that innovatively incorporates emergency rescue functions into traditional physiological monitoring systems using the rescue of modified-MAC (RM-MAC) protocol. The relay selection (RS) algorithm of APMRS can select the appropriate relay to forward based on the importance of the physiological data, thus ensuring priority transmission of more important monitoring data. In addition, we apply deep learning target trajectory prediction technology to the indoor rescue module (IRM) of APMRS to provide high-performance scheduling of location tracking nodes in advance by trajectory prediction. It reduces network energy consumption and ensures perceptual tracking accuracy. When APMRS monitors abnormal physiological data that may endanger a patient's life, IRM can implement effective and fast location rescue to avoid risks.
Keywords: deep learning node scheduling; real-time physiological monitoring; risk avoidance and rescue; wireless wearable biosensors.