Acute respiratory distress syndrome (ARDS) is a critical lung condition caused by trauma or infection. This study explores the development and evaluation of human lung phantoms to investigate the feasibility of using microwave frequencies for ARDS detection. Both physical semisolid phantoms and their numerical models were developed in inflated and deflated states to replicate the dielectric properties of healthy and affected lungs. Three phantom sets with varying water and air content were fabricated to simulate different stages of respiratory distress. The geometric parameters of the phantoms were derived from CT scans of 166 ARDS patients. Dielectric permittivity and conductivity were measured using a Keysight N1501A dielectric probe over a 0.5–13 GHz range, showing strong agreement with IFAC’s reference data. To validate the models, horn antennas operating between 8.2–12.4 GHz were used to measure S-parameters (S11 and S21) in both physical and numerical phantoms. The results demonstrated consistent changes in transmission and reflection characteristics corresponding to variations in lung volume and dielectric properties. These findings support the potential of microwave imaging as a non-invasive tool for early ARDS detection by effectively distinguishing between healthy and distressed lung states based on measurable electromagnetic response.