Comprehensive Summary
This paper discusses the computational optimization of 3D-printed PLA+ lattice scaffolds for bone tissue engineering using an integrated methodology of Taguchi L_27 experimental design, Back-propagation Artificial Neural Network (BPANN), and Finite Element Analysis (FEA). Three TPMS geometries (Lidinoid, Gyroid, Diamond), wall thicknesses (1.0 mm, 1.5 mm, 2.0 mm), and compressive loads (3 kN, 6 kN, 9 kN) were tested, with displacement and strain as outputs. ANOVA identified geometry as the main determinant of mechanical response, contributing to 84.38% of displacement variation, followed by wall thickness (11.37%) and applied load (1.56%). Gyroid geometries demonstrated the least displacement and strain at higher wall thicknesses, Diamond exhibited balanced performance, and Lidinoid showed the highest deformability. Distinct geometry failure was also observed, with Lidinoid failing by buckling, Diamond by shear, and Gyroid by ductile-like, energy absorbing collapse. In addition, increasing wall thickness demonstrated enhanced stiffness and reduced displacement and strain across all geometries. The BPANN model was trained on experimental data and had high predictive accuracy (R^2 = 0.9991 for displacement, 0.9954 for strain; MAPE = 0.79%), while FEA analysis matched experimental trends within 6 - 17% error. These findings underline Gyroid scaffolds, especially at 2.0 mm thickness, as the most promising design given its stiffness, uniform stress distribution, and stable collapse failure, while highlighting hybrid scaffolds balancing the strengths of each geometry as a way to further optimize porosity, stiffness, strength, and stress distribution. Such factors directly influence osteointegration, cell proliferation, nutrient transport, and vascularization, all critical for successful tissue regeneration and long-term stability. The authors further note that future studies should focus on the refinement of 3D printing parameters (infill density, build direction, print resolution) as well as scaffold design optimization to guarantee consistency, structural integrity, and biocompatibility.
Outcomes and Implications
This framework demonstrates how integrating computational tools into experimental methodology can jointly advance scaffold optimization. BPANN enables efficient pre-clinical design and FEA can validate experimental and predicted results, together guiding the design of patient-specific scaffolds tailored to defect size and loading environment. Clinically, such scaffolds could address challenging cases of non-union fractures where bones fail to heal due to bone defects or loss from trauma, diseases, or congenital abnormalities. More broadly, lattice-based scaffolds designed with this optimized approach could accelerate bone tissue regeneration, improve vascularization, and reduce implant failures, providing a pathway towards scaffolds that balance mechanical integrity with biological compatibility.