• The ensemble surrogate model, integrating Kriging and radial basis function models, significantly improves prediction accuracy for coronary stent mechanical properties under limited sample sizes.
• The proposed method achieves dual-objective optimization of stent radial stiffness and bending stiffness, outperforming single surrogate models.
• The ensemble surrogate model reduces the inverse radial stiffness by up to 13.92% and improves bending stiffness by up to 2.56% compared to single models.
• The framework offers a low-cost, high-precision solution for coronary stent design, addressing the trade-off between support and flexibility.