Temperature-induced curling and moisture-driven warping continually reshape jointed concrete pavements (JCP), altering wheel-path profiles and affecting the International Roughness Index (IRI). This report presents an integrated framework combining finite element method (FEM) analyses, artificial neural networks (ANNs), and IRI modeling to isolate and quantify the roughness component caused by slab curvature. A large factorial set of ISLAB2000 simulations covering slab thickness, joint spacing, subgrade stiffness, joint moment stiffness, and equivalent total temperature gradient provides mechanistic responses for deflection, curvature, and longitudinal stress used to train ANNs. Synthetic 500-ft profiles derived from these deflections are analyzed with a quarter-car model to compute curvature-only IRI and perform sensitivity studies. Results indicate that IRI is strongly governed by the temperature gradient, joint-moment stiffness, and slab geometry, with a linear relationship between average curvature and IRI. A back-calculation procedure applied to field profiles from three sites estimates equivalent total temperature gradients and curvature-only IRI, indicating that a significant portion of the measured IRI is attributable to diurnal curling and warping. The framework is implemented in the SLAB CURL graphical user interface, providing a practical tool for relating slab curvature, ΔT/h, and IRI for design- and network-level JCP evaluation.