Artificial intelligence investigation of three silicates bioceramics-magnetite bio-nanocomposite: Hyperthermia and biomedical applications

• Published in: Nanomedicine journal Vol. 5; no. 3; pp. 163 - 171
• Main Authors: Amir Hussein Montazeran, Saeed Saber Samandari, Amirsalar Khandan
• Format: Journal Article
• Language: English
• Published: Mashhad University of Medical Sciences 01.07.2018
• Subjects: Artificial neural network; Biomedicine; Magnetite nanoparticle; Nanocomposite; Tissue engineering
• ISSN: 2322-3049; 2322-5904
• DOI: 10.22038/nmj.2018.005.0006

Objective(s): Bioactive silicate ceramics have favorable features for applying as off-the-shelf bone and artificial tissue. Calcium silicate can enhance the generation of an immediate bond with host bone without an intervening rough surface in the bone layer. However, the silicate bioceramics have some drawback regarding their mechanical properties and chemical stabilities. Materials and Methods: In this study, magnetite nanoparticles (MNPs) as reinforcement were added to the three silicate bioceramics to investigate the physical and mechanical properties as well as their magnetic behavior as a case study and compare with other calcium silicate nanocomposite which are excellent candidates for hyperthermia applications. Then the artificial neural network (ANN) applied to the previous data to predict the mechanical and biological behavior of the bio-nanocomposite as output parameters. A predicted model was enhanced using ANN to measure the optimum size and reinforcement amount of the magnetite bio-nanocomposite. The results of the fabricated bio-nanocomposite were extracted experimentally corresponding to different MNPs weight fractions compared to the predicted model. Results: The X-ray diffraction (XRD), scan electron microscopy (SEM) technique were used to compare the porosity and porous tissue microstructure. Thereafter, an analytical solution is presented to express explicitly the physical and mechanical responses of the bulk/scaffold bio-nanocomposite. Conclusion: The obtained results showed the potential application of these calculations and analyses in a wide range of numerical studies. The comparison presented within the test and predicted values showed that the modeling outcomes were close to testing values.