FEM-based Parametric Optimization of a Measurement Setup for Sensitivity Improvement in Insulin Absorption Assessment

A finite element model of human abdominal tissue for improving the percentage relative sensitivity of a bioimpedance spectroscopy-based method for insulin absorption measurements was proposed. The model was realised by considering only four layers of the abdomen (dry and wet skin, fat, and muscle) electrically characterized with models already validated in the scientific literature. A parametric optimization was carried out to maximize the relative sensitivity of the proposed measurement method. In particular, a fractional factorial plan approach (Taguchi L18) was implemented to analyze different setups of the measurement apparatus in terms of signal frequency and dimensions and electrodes mutual distance. ANOM and ANOVA were conducted to assess the impact of each parameter on the measurement percentage relative sensitivity. The analysis revealed that the distance between the amperometric and voltammetric electrode as the most influencing factor on the percentage relative sensitivity. Considering the constraints related to the direction of the total current density field and the electrodes positioning, the value of the distance maximizing the sensitivity results the optimum one. A range of percentage relative sensitivity variation of 29.25 % was obtained in the experimental plan investigated. The highest percentage relative sensitivity of 29.78 % was obtained with the optimum measurement setup, in the case of 100 ml insulin injected. An improvement in percentage relative sensitivity by a factor of 30 was achieved with the optimum measurement setup with respect to the measurement setup proposed in literature.