<p>The performance and sensitivity of microcapacitive pressure sensors are essential for high-precision biomedical and ultrasonic applications. This study introduces a novel multi-physics theoretical framework that integrates nonlinear poroelastic behavior, large-deflection plate mechanics, and porosity-dependent dielectric properties to model capacitive sensors with soft porous polymeric dielectrics, overcoming limitations of conventional air-gap models. The finite element-based Galerkin method is employed to solve the coupled nonlinear governing equations. Four porous polydimethylsiloxane (PDMS) variants with tailored Young’s moduli and permittivities are analyzed and compared to a vacuum dielectric baseline. Key novelties include: (i) a physics-based model capturing dynamic compression and permittivity evolution in porous dielectrics, (ii) a free-standing upper electrode design enabled by soft dielectrics, eliminating sealed cavities, and (iii) a hybrid machine learning (ANN)-augmented approach for rapid performance prediction and inverse material design. Results show that low-stiffness, high-permittivity polymers significantly enhance sensitivity (up to 10 times), reduce actuation voltage, and increase resonant frequency critical for CMUTs and wearable sensors. The ANN model achieves R² &gt; 0.97 in predicting sensitivity, pull-in voltage, and resonance frequency, enabling efficient optimization. Simulation results are validated against COMSOL Multiphysics 6.1. This work establishes a generalized design platform for next-generation high-performance capacitive MEMS.</p>