Explainable Artificial Intelligence‐Integrated Optimization and Experimental Validation of Mechanical Performance in Mg‐Based Composites

The development of magnesium‐based metal matrix composites (MMCs) for biomedical implants is often hindered by the high cost and time associated with conventional trial‐and‐error optimization. In this study, we present an explainable ML framework to accelerate the design and prediction of mechanical performance in magnesium‐based alloys reinforced with HAP and a hybrid system of HAP and GNP. An ensemble of regression models was evaluated, with XGBoost demonstrating superior generalization capability, achieving R ² values of 0.955 for yield strength, 0.954 for ultimate tensile strength, and 0.939 for % elongation. To ensure model transparency, SHAP was integrated to interpret feature contributions at both global and local levels. The analysis revealed that ECAP and weight percentage of reinforcement are the most influential factors, with nonlinear optimal ranges identified for weight percentage of reinforcement, particle size, and reinforcement. Guided by these insights, a new AZ91‐matrix composite incorporating a hybrid HAP and GNP reinforcement was designed and fabricated via stir casting. Experimental validation demonstrated excellent agreement between model predictions and measured mechanical responses, confirming the high‐fidelity framework. Residual analysis confirmed model robustness and homoscedasticity. This work establishes a transparent, data‐driven framework that bridges predictive modeling with physical interpretability, enabling the rational design of high‐performance, biodegradable implants with reduced reliance on extensive experimentation.