Diabetic foot is a common complication in patients with diabetes, which can lead to plantar ulcers and even necessitate amputation. This study aims to utilize finite element analysis to simulate the offloading effects of 3D-printed insoles with various structures on plantar pressure and to explore the use of machine learning in providing optimal plantar pressure offloading solutions for patients with diabetic foot. The results demonstrated that negative Poisson’s ratio structured insoles were more effective in reducing plantar pressure (reducing pressure by an average of 39.2%) than barefoot and conventional structures. This was achieved through a unique lateral contraction deformation, which increased the contact area with the foot. The pressure-reducing effect of insoles may be weight-related, suggesting that heavier patients may require stiffer insoles. However, the machine learning algorithm demonstrated a poor fit (only 60.75%) in the task of recommending suitable insoles. In conclusion, this study demonstrated the significant effect of negative Poisson’s ratio structured insoles in reducing plantar pressure in diabetic patients, providing new ideas for diabetic foot protection. With the development of data analysis technology in the future, the feasibility and application of personalised insole design will be more promising.