City Research Online

Abstract

Electrical impedance measurements on finite samples are affected by geometric factors, causing deviations from ideal Ohm's law. Finite Element Modeling (FEM) is typically used to study these effects, but it's computationally intensive. This work presents a faster, regression-based approach to predict the resistance of L×W×H samples at a fixed frequency. A dataset was generated using COMSOL Multiphysics for samples with constant conductivity, where geometry (L,W,H) was systematically varied. Initially, a complex cubic polynomial regression model was developed, achieving high accuracy (R² = 0.9999, Mean Absolute Percentage Error MAPE = 0.2751%) but generating 44 terms. To increase practicality, an optimized Feature-Engineered model was created by eliminating coefficients with high p-value, retaining only 1 intercept and 10 coefficients. This simplified model maintained good predictive capability (R² = 0.9993, MAPE = 1.3394%), making it computationally efficient. Results demonstrate that simple, optimized regression models are an efficient, rapid alternative to FEM for geometry-dependent resistance prediction.

Publication Type:	Conference or Workshop Item (Paper)
Additional Information:	© 2026 IEEE. This accepted manuscript is made available under the terms of the Creative Commons Attribution License (CC-BY), which permits unrestricted use, distribution and reproduction in any medium, provided the original work is properly cited.
Publisher Keywords:	Bio-impedance Spectroscopy, Finite Element Modeling, Regression Modeling, COMSOL Multiphysics Simulation, Resistance Prediction, Optimization
Departments:	School of Health & Medical Sciences School of Health & Medical Sciences > Department of Optometry & Visual Science School of Science & Technology School of Science & Technology > Department of Engineering
SWORD Depositor:	Symplectic Administrator

Export

Downloads