Through spatial arrangement of different microstructures, well-designed hybrid lattice structures exhibit enhanced mechanical properties compared with uniform lattice structures composed of a single unit cell, thereby serving as a promising strategy for lightweight lattice structure design. However, current microstructure design approaches for hybrid lattices primarily focus on improving quasi-static compressive mechanical performance. In reality, engineering protection often involves complex dynamic impact environments rather than simple quasi-static loads. However, research on the microstructural design of hybrid lattices under high strain rate impact loading remains limited. This is due to the significant influence of inertial effects on dynamic structural mechanical response, which complicates multi-objective optimization. A fourth-order hybrid orthogonal isotropic lattice structure was constructed using two types of rhombic dodecahedron (RD) unit cells with different relative densities. A bidirectional long short-term memory (Bi-LSTM) network was employed to develop a predictive model that expanded the dataset of stress-strain curves obtained from finite element simulations. The reliability of the model’s predictions was validated through multi-dimensional evaluation metrics based on the predicted stress points along the stress–strain curves. The coefficient of determination (R²) of the predicted results reached 0.971 and 0.990 under dynamic and quasi-static compression, respectively. Specific energy absorption (SEA) was adopted as an indicator for evaluating energy absorption capacity based on the accurately predicted stress-strain curves. The structural effects on the energy absorption behavior of the hybrid lattice were analyzed under quasi-static planar compression. By increasing the compression speed to 100 m/s to simulate high strain rate conditions, the role of inertial effects on the dynamic mechanical response was investigated through comparative analysis. Furthermore, the competition and synergy between inertial effects and structural effects were systematically analyzed by categorizing the hybrid lattices according to the degree of unit cell mixing. The results indicate that inertial effects become dominant under high strain rate loading, while structural effects govern the response under quasi-static conditions. By using machine learning to efficiently expand the data scale of finite element simulations, the internal relationships among inertial effects, unit cell proportion, and spatial arrangement under high strain rate conditions were revealed. The findings provide valuable insights for the application of hybrid lattice structures in impact protection and related engineering fields.