Content area
Abstract
Over the last few years, there has been a growing interest in the development of biodegradable implants that can provide temporary support until bones heal and degrade gradually, eliminating the need for additional surgical interventions. Magnesium (Mg) alloys have emerged as promising candidates for biodegradable implants due to their biocompatible and biodegradable properties, as well as their physical characteristics similar to human bone, which reduce stress shielding during load transfer at the bone-implant interface. However, the low mechanical strength of these alloys remains a significant challenge, limiting their use as bone substitutes and cardiovascular stents. Optimizing the mechanical properties of biocompatible Mg alloys through experimental methods can be time-consuming and costly. Therefore, machine learning models offer a unique advantage in understanding the complex relationship between different alloy features and mechanical properties. In this study, a three-step method was proposed to design new alloys. In the first step, various machine learning models were used to analyze the relationship between microstructure, composition, heat treatment, and mechanical properties of biocompatible Mg alloys. A dataset of literature data and thermodynamics simulation data from FactSage ⃝c software were utilized to conduct the analyses. The results demonstrated that the random forest model was the most accurate model for predicting ultimate tensile strength (UTS) and yield strength (YS) with 90% and 91% R2, respectively. The use of thermodynamic data, along with appropriate data preprocessing techniques, resulted in acceptable model accuracy. In the second step, the model was validated through CALPHAD technique simulations, and good agreement was observed between the model predictions and the results obtained from the simulations. In the final step, the genetic algorithm was employed in conjunction with the random forest model to identify the most promising candidate alloys for use as biodegradable implants with optimized composition to have the highest mechanical properties. Following this methodology, the proposed candidate alloys were synthesized, cast, and tested, offering high strength, biocompatibility, and biodegradability. The use of machine learning in combination with thermodynamic simulations and genetic algorithms demonstrates the feasibility of designing advanced biodegradable implants with optimized mechanical properties. By utilizing this methodology, researchers can efficiently design and test new alloys, potentially accelerating the development of biodegradable implants and improving patient outcomes.





