Application of artificial neural networks to predict Young's moduli of cartilage scaffolds: An in-vitro and micromechanical study

In this study, four-phase Gelatin-Polypyrrole-Akermanite-Magnetite scaffolds were fabricated and analyzed using in-vitro tests and numerical simulations. Such scaffolds contained various amounts of Magnetite bioceramics as much as 0, 5, 10, and 15 wt% of Gelatin-Polypyrrole-Akermanite biocomposite. X-ray diffraction analysis and Fourier transform infrared spectroscopy were conducted. Swelling and degradation of the scaffolds were studied by immersing them in phosphate-buffered saline, PBS, solution. Magnetite bioceramics decreased the swelling percent and degradation duration. By immersing scaffolds in simulated body fluid, the highest formation rate of Apatite was observed in the 15 wt% Magnetite samples. The mean pore size was in an acceptable range to provide suitable conditions for cell proliferation. MG-63 cells were cultured on extracts of the scaffolds for 24, 48, and 72 h and their surfaces for 24 h. Cell viabilities and cell morphologies were assessed. Afterward, micromechanical models with spherical and polyhedral voids and artificial neural networks were employed to predict Young's moduli of the scaffolds. Based on the results of finite element analyses, spherical-shaped void models made the best predictions of elastic behavior in the 0, 5 wt% Magnetite scaffolds compared to the experimental data. Results of the simulations and experimental tests for the ten wt% Magnetite samples were well matched in both micromechanical models. In the 15 wt% Magnetite sample, models with polyhedral voids could precisely predict Young's modulus of such scaffolds.