Solid though hydrogels are a class of bio-inspired elastomeric material with the ability of high stretch-ability, toughness, and water absorption. Though hydrogels exhibit nonlinear features such as stress softening, permanent set, cyclic energy dissipation and induced anisotropy in their mechanical responses. Besides, gels used in biomedical devices and soft robots are designed to sustain continuous cyclic loading during their service-life. While most studies assume that the majority of stress softening occurs in the first cycle. However, experimental studies suggest that the material exhibits damage accumulation and stiffness reduction significantly at higher number of cyclic loading. This progressive stress softening can lead to an unpredicted response of the material in dynamic loading as it is a function of material stiffness. However, there are only a few constitutive models developed to address this accumulative damage in the elastomeric gels. Thus, here, we developed a data-driven method to predict the behavior of hydrogels subjected to a high number of cycles. The developed model can be particularly used to estimate material response in each cycle and its steady-state response due to three-dimensional deformation. The strain energy of the material is the same in all direction in the undeformed state and damage evolve in each direction based on its maximum deformation. Moreover, the energy of the material in each direction estimated through two parallel artificial neural network. The proposed model shows good agreement with cyclic uni-axial tensile test data in different cycles of the loading and steady-state response of material.