Modeling and simulation of human lung mechanics using single to multi-compartment electrical circuit models

Abstract

This study focuses on the development, simulation, and comparative analysis of electrical analog models for representing human respiratory mechanics. The respiratory system was modeled using lumped parameter elec¬trical circuits, where physiological variables such as pressure, airflow, lung compliance, airway resistance, and airway inertance were represented by voltage, current, capacitance, resistance, and inductance, respectively. Four distinct models: single-compartment, two-compartment, three-compartment, and multi-compartment were developed and implemented using MATLAB/Simulink and to investigate their predictive accuracy and com¬putational efficiency under both normal and diseased lung conditions, including asthma, COPD, and ARDS. Performance was evaluated using Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and correla¬tion coefficients by comparing the single-, two-, and three-compartment model outputs against a high-fidelity ten-compartment (𝑁𝑁 𝑁 𝑁𝑁) reference model, rather than against patient or clinical ventilator data. Results indi¬cated that predictive error decreased, and correlation with the reference model increased, monotonically with compartment count (from a correlation coefficient of 0.9990 for the single-compartment model to 1.0000 for the 𝑁𝑁 𝑁 𝑁𝑁 reference model itself); the multi-compartment model’s computational complexity makes it more suitable for offline analysis. A weighted decision matrix was used to compare the single-, two-, and three-compartment models on predictive accuracy, computational speed, clinical applicability, ease of parameter estimation, and model stability. A sensitivity analysis showed that the top-ranked model depends materially on how the qualitative criteria (clinical applicability, parameter estimation ease, stability) are scored: under an illustrative expert-judgment scoring the three-compartment model scored highest, while under a fully formula-derived scoring the two-compartment model scored highest instead, and the single-compartment model became competitive with both under weight assumptions with no built-in preference. The study therefore identifies the two- and three-compartment models as the strongest real-time candidates, with the specific choice between them depending on which criteria a given clinical deployment prioritizes, rather than concluding that a single configuration is unconditionally optimal.