Accurate in-situ stress characterization is essential for hydraulic fracturing design in shale reservoirs, yet 3D geomechanical model calibration still relies on iterative trial-and-error procedures that are computationally intensive and physically opaque. Sensitivity analysis, though widely used to evaluate the influence of elastic parameters, is typically treated as descriptive rather than predictive. This work proposes a sensitivity-driven inversion framework for efficient calibration of horizontal stresses in 3D geomechanical models. Controlled perturbation simulations on a baseline model quantify the responses of maximum and minimum horizontal stresses ($S_H$ and $S_h$) to variations in... More >