The cushioning performance of insole has always been a critical consideration in its design. While the development of intelligent methods and the emergence of additive manufacturing (AM) technology have enhanced design freedom and convenience, a standardized approach to guide designers in selecting optimal materials and structures for specific scenarios is still lacking. This study aims to propose a controllable parameterized lattice cushioning insole (PLI) by integrating finite element (FE) and machine learning (ML) methods. The insole performance can be adjusted by modifying the structural parameters (a, b) and the internal strut thickness (t). The findings indicate that PLI, under the optimal parameter combination (a = 2.54, b = 3.56, t = 3.15), can reduce plantar pressure by up to 44.45 %, which may be achieved by increasing the contact between the footwear and the foot. The data-driven PLI optimization design method proposed in this study significantly enhances the cushioning performance of insole structures, simplifies the optimization process for selecting insole structures or materials, and provides a systematic and efficient solution for insole design. Although the initial preparation of material data is time-intensive, the trained model eliminates the need for repeated laboratory gait analysis or plantar pressure measurements, offering a foundational reference for clinical applications in insole structure design.