CORROSION PREDICTION OFMAGNESIUM IMPLANT USING MULTISCALE MODELING BASED ON MACHINE LEARNING ALGORITHMS

Abstract

Significant thoughtful research is really necessary to improve the patient outcomes and reduce the social and financial burdens associated with implant failure. The primary focus of the researchers is to minimize the major implant failure due to corrosion attributed to making orthopedic surgery safer and more effective. Hence, a critical review has been done in this present article on the various multiscale modelings based on machine learning algorithms (MLAs) to predict the corrosion behavior of magnesium (Mg) alloy implants. According to the best of the authors' knowledge, all the available multiscale modelings tools, such as artificial neural network (ANN), least absolute shrinkage and selection operator (LASSO) regression model, multiple linear regression and random forest regression (RFR) models, etc., are methodically presented and discussed in detailed here for the prediction of corrosion mechanism. Subsequently, various multiscale model tools and assessment metrics for models have been thoroughly compared and criticized for better understanding and optimizing of the corrosion behavior of implants. The comparison indicates that the RFR model may be the best option, whereas the LASSO regression model and ANNs show inefficient performance for the prediction of corrosion behavior. Apart from the multiscale modeling approach, the authors have also explored the physiology and properties of alloys, bone implant, immune and tissue system, and the corrosion control mechanisms of Mg alloy. Finally, the present review on multiscale modeling approach and assessment metrics models will enhance the knowledge and understanding of the corrosion behavior of Mg alloy for implant application.