Enhancing inter-subject reproducibility in insulin bioavailability measurements through real-time calibration

Abstract

A real-time calibration to enhance the inter-subject reproducibility of insulin bioavailability measurements was proposed and validated on a diabetic patient with more than 49000 impedance values. The measurement method monitors drug absorption through a transducer consisting of a sensitive material (human abdominal tissue) and an impedance spectrometer. The in-vivo experiments revealed a statistically significant negative second-order polynomial trend in impedance magnitude corresponding to the amount of rapid-acting insulin injected. To manage the uncertainty inherent in the sensing block, the method integrates real-time calibration. Implementing a real-time calibration method based on a customized model identification at each insulin administration resulted in a reduction in deterministic error by a factor of 18. A further reduction in deterministic error for clinical application (46%) was achieved with a customized parameter second-order polynomial model with respect to the customized parameter linear model. Lastly, when less conductive long-acting insulin was administered, an increasing trend in impedance magnitude was observed, aligned with the theoretical framework of the proposed method.