Computational optimization of 3D printed bone scaffolds using orthogonal array-driven FEA and neural network modeling

Today, orthopedic surgeons have been continuously focusing on bone tissue engineering for regenerating damaged bone through the use of biomimetic scaffolds and innovative materials. Hence, this study presents a comprehensive investigation into the optimization of PLA + 3D printed lattice scaffolds for bone tissue engineering applications, emphasizing the role of geometric configuration and processing parameters on mechanical performance. Three distinct lattice geometries such as Lidinoid, Diamond, and Gyroid were developed with varying wall thicknesses (1.0 mm, 1.5 mm, and 2.0 mm) and subjected to compressive loads of 3 kN, 6 kN, and 9 kN. A Taguchi L₂₇ Orthogonal Array was employed to evaluate key mechanical responses, including displacement and strain. Among these configurations, the Gyroid lattice exhibited superior mechanical integrity, demonstrating the least displacement (0.36 mm) and strain (1.2 × 10⁻²) at 3 kN with 2.0 mm thickness, whereas the Lidinoid structure showed the highest deformability. A Back-propagation Artificial Neural Network (BPANN) model was developed to predict scaffold behavior with remarkable accuracy (R² = 0.9991 for displacement, R² = 0.9954 for strain), further Finite Element Analysis (FEA) was conducted to validate both experimental and predicted results. The novelty of this work lies in its integrative, multi-modal approach that synergizes experimental design, machine learning-based predictive modeling, and simulation. The focus of this study is to define a robust framework for optimizing scaffold architecture, with significant implications for enhancing mechanical strength and biological performance in bone healing applications.