The compensation of perturbing temperature fluctuation in glucose monitoring technologies based on impedance spectroscopy

Non-invasive glucose monitoring techniques based on impedance spectroscopy are affected by a variety of perturbing effects. In order to use the impedance as a glucose measure, these perturbing effects need to be quantified and compensated. Since effects induced by temperature fluctuations certainly rank among the severest perturbations, a clinical study was carried out to establish whether temperature, as a perturbing factor, could be compensated for in impedance spectroscopy. The results as well as a concept allowing for the compensation of perturbing temperature fluctuations are presented here. The compensation technique described is a generic approach that, in principle, can be applied to compensate most perturbation effects provided that there are now multiplicative interactions between the variable of interest (in our case the glucose) and the perturbations. The results allow for the determination of the minimum required sensitivity of an impedance spectrometer to glucose in order to be operational in home-use conditions. Furthermore, the data can be used to estimate if a universal temperature compensation can be applied or if an individual calibration is necessary. For instance, applying a universal temperature compensation and requiring an application range of ±5°C, the minimum required sensitivity of the minimum impedance and frequency in a sensor-skin RLC circuit to resolve glucose variations equivalent to 10 mg/dl is 0.85 Ω/mg/dl and 0.14 MHz/mg/dl, respectively. The sensitivity requirements reduce by about a factor 1.6, if for each subject an individual calibration is carried out. Depending on the measure and the calibration procedure, the required sensitivities are a factor 3–50 greater than those reported in the literature. Thus, in order to be operational in home-use conditions, the signal-to-noise ratio (S/N) of existing impedance-based monitoring platforms using RLC circuits need to be improved by about one order of magnitude. In order to make non-invasive glucose monitoring possible we, therefore, suggest some measures that may improve the S/N by the required factor.