The performance and durability of proton exchange membrane fuel cells (PEMFCs) are strongly governed by the electrical contact resistance (ECR) at the gas diffusion layer (GDL)–bipolar plate (BPP) interface, where up to 59% of total power losses can originate. ECR arises from the complex coupling between GDL microstructure and interfacial contact mechanics, underscoring the need for accurate, physics-informed characterization. This work introduces an integrative framework that combines a physics-based electrical contact model with machine learning (ML) techniques to identify and quantify the influence of morphological and mechanical descriptors of carbon paper GDLs on ECR. A comprehensive, pressure-dependent dataset is generated by simulating GDL microstructures using a Poisson line process under diverse configurations, while the BPP is idealized as a perfectly smooth surface. ECR is computed using a Holm–Greenwood–type model that represents interfacial contacts via ensembles of a-spots, each approximated by circles whose areas match the real GDL–BPP contact regions. To address the geometric simplifications inherent in this approach, we systematically compare alternative digitalizations of fiber–plate contact—including elliptical, rectangular, and circular approximations—and evaluate their convergence toward analytical ECR predictions. The resulting framework provides a robust pathway for elucidating structure–property relationships in GDLs and for guiding the design of low-resistance PEMFC interfaces.