Mechanobiological design strategy for additively manufacturable tibial ankle implant for enhanced biomechanical and osseointegration performance: A finite element and machine learning approach

The leading cause of revision surgeries in ankle arthroplasty is aseptic loosening of the tibial implant, resulting from adverse bone remodelling and insufficient osseointegration. Aseptic loosening depends on multiple factors, such as the design of the implant, the porous surface of the implant, the quality of bone, implant positioning, wear debris, etc. The extent to which the design of porous architecture and its relationship with aseptic loosening failure mechanisms remains unexplored. The study aims to identify the lattice design of porous rhombic dodecahedron architecture of tibial implants that would be able to maximise bone formation and reduce stress shielding using macro-micro finite element (FE) analysis with machine learning (ML) approach. The study entails the macro-microscale FE modelling of four porous rhombic dodecahedron tibial implants (PRDTI), referred as PRDTI50, PRDTI60, PRDTI70, and PRDTI80. Based on macro-micro-FE determined dataset, four artificial neural network (ANN)-based ML algorithms were trained and validated for faster prediction of bone ingrowth. Results evidenced that von Mises stress in the tibia exhibited elevated stresses for PRDTI80 and PRDTI70 implants compared to PRDTI60, PRDTI50, and solid implant. Bone ingrowth results indicated that the PRDTI70 implant exhibited higher amounts of bone formation. The study proposes the PRDTI70 implant is a viable option for designing tibial implants to simultaneously reduce stress shielding and maximises bone ingrowth. This preclinical analysis sheds light on the role of porous structure design in bone formation for the development of porous tibial prostheses for TAR to prevent revision instances.