Auslegung eines Sensorsystems zur Überwachung der menschlichen Hydratation

In this research project, a concept for the evaluation of a mobile sensor system is developed, which contributes to the non-invasive and continuous measurement of the hydration status. Through targeted preliminary investigations of tissue moisture and the determination of hydration status, an algorithm is established which allows a conclusion to be drawn about the total body water of the human being by means of sensor data fusion of various vital signs. The concept for monitoring the hydration state is based on continuous vital signs monitoring of the parameters pulse, oxygen saturation, bioimpedance, skin conductance and tissue hydration. By means of person-specific overnight calibration, the system is adjusted to the user. The calibration is based on an extensive correlation analysis between the collected vital signs and weight, which can be considered representative of hydration status. The algorithm is validated with the help of data acquisition from 9 participants. Data from two nights were collected for calibration. The correlation analysis shows noticeably that the collected parameters run in opposite directions during the night as well as the preliminary tests. This behaviour is largely due to sleep phases and other environmental conditions during sleep. By acquiring more data during the day, the calibrated algorithm was tested. Verification that the algorithm calculates valid values of total body water is performed by steady weight control. Through correlation analysis of the day data sets, it can be determined that the parameters are again trending in the same direction as originally determined in the preliminary tests. The results of the correlation analysis of the night records lead to an incorrect calibration of the algorithm. Regarding this, a different way of calibrating the system must be chosen and the algorithm needs to be revalidated. Actual dehydration can be detected by the system in only 33.3% of the participants. This shows that it is in principle possible to detect dehydration by continuous vital signs monitoring and subsequent sensor data fusion. Calibration and accounting of the data will also be available on the device so that the algorithm can be applied in real time. A calibration process during the day is aimed at; a corresponding concept for this must be developed and evaluated in the future.