Recently, a notable breakthrough has been achieved in the field of data-driven inverse design for multifunctional bicontinuous multiscale structures. This advancement comes from the collaborative efforts of Professor Liu Ligang, Associate Professor Fu Xiaoming, and Associate Researcher Zhai Xiaoya, all from the School of Mathematical Sciences at the University of Science and Technology of China (USTC). They worked in tandem with teams from Xinjiang Normal University, the Zhongguancun Artificial Intelligence Research Institute, and The Hong Kong Polytechnic University.

Their significant research findings were published in Nature Communications on January 8, 2026, under the title "Data-driven Inverse Design of Multifunctional Bicontinuous Multiscale Structures." This study represents the first systematic approach to addressing the long-standing core bottlenecks of bicontinuous multiscale structures, namely the "difficulty in description, design, and manufacturing." It introduces a groundbreaking data-driven paradigm for the intelligent design of complex engineering systems, including bone implants, permeable devices, and mechanical stealth structures.

The research team introduced an innovative data-driven inverse design method tailored for bicontinuous multiscale structures. By establishing the L-BOM dataset and combining active learning with generative artificial intelligence models, they created a closed-loop design framework encompassing "generation, screening, and retraining." This approach significantly broadens the designable performance space.

Taking femoral implants as a case study, the design outcomes closely mirrored natural bone tissue in crucial metrics like Young's modulus, pore size, and porosity, achieving an exact match between structural performance and biological characteristics. Furthermore, the team designed and verified a bicontinuous multiscale filtration structure that outperformed traditional structures in terms of permeability and specific surface area.

This research presents a novel data-driven paradigm for the intelligent inverse design of multifunctional, multi-physics structural materials. It boosts the design efficiency of complex structures by over two orders of magnitude, establishing a vital foundation for engineering applications such as customized bone implants, permeable devices, and mechanical stealth structures.