Biomechanical design of titanium-PEEK combined fusion cage based on PLIF surgical model

Fusion devices play a critical role in lumbar fusion surgery. Titanium alloy fusion devices offer good biocompatibility and stability, but their mechanical properties far exceed those of bone, leading to stress shielding effects after implantation, which can reduce spinal fusion rates and cause endplate collapse. On the other hand, fusion devices made of polyether ether ketone (PEEK), which has a lower elastic modulus, are not conducive to bone ingrowth and fusion stability due to their material properties. Personalized fusion devices that can precisely adapt to a patient's physiological condition are not widely used due to their lengthy design cycle. This study proposes an optimized design method based on a titanium alloy-PEEK composite structure. By constructing three composite structure models-PEEK core and Ti frame (square hole type, circular hole type, plate type)-and combining finite element compression simulation with machine learning algorithms, the structural parameters are intelligently optimized. The machine learning algorithm used in this study is Back Propagation Neural Network. The aim of this study is to match the equivalent elastic modulus of the fusion device with that of cortical bone. The three optimized fusion devices, along with the Ti fusion device and PEEK fusion device as control groups, were implanted into a traditional PLIF postoperative model for static and transient dynamic analysis. The biomechanical responses of the lumbar spine at various locations after implantation of the five fusion devices were analyzed and compared. The results indicate that all three optimized fusion devices effectively reduce the risk of device settlement, thereby mitigating stress shielding effects, improving fusion rates, and enhancing postoperative lumbar stability. Among them, the circular hole inner core fusion device (M2) demonstrated the best overall performance. The peak von Mises stress of L4 lower endplate and L5 upper endplate in M2 model were 54.2% and 27.7% respectively lower than those in Ti fusion device. Compared with Ti fusion device, the strain energy of M2 model increased by 49.7%. The development framework of this study which integrated "finite element simulation-machine learning-postoperative model biomechanical validation and evaluation" can effectively reduce the design cycle and cost of personalized orthopedic implants.