Journal of Alloys and Compounds, vol.745, pp.840-848, 2018 (SCI-Expanded)
Using artificial intelligence (AI) applications such as, Artificial neural network (ANN) and Gene expression programming (GEP) to model corrosion performance of the hydroxyapatite coated metallic biomaterials were performed. Created models were analyzed and compared with a response surface methodology (RSM) study. Electrodeposition parameters of the hydroxyapatite on CoCrMo implant materials were used as independent variables in the modeling study. Corrosion potential, Ecorr values calculated from potentiodynamic polarization measurements were used as dependent variable (output data) in AI models. Effect of the deposition parameters on the in vitro corrosion performance of the hydroxyapatite coatings were modeled by ANN and GEP models. ANN models were built with 3 input and 1 output variables using Multilayer Perception Topology. Total 21 electrodeposition and corrosion experiments were used in the ANN and GEP modeling. AI applications were successful for modeling the effect of deposition parameters on the corrosion performance of the hydroxyapatite coatings. Individual effect of the each parameter was investigated statistically. According to the model results predictive capacity and effectiveness of the ANN model is slightly better compared to the GEP and RSM model.