A computationally efficient stochastic analysis method for predicting the long-term development of aneurysms

The long-term progression and rupture mechanisms of aneurysms remain poorly understood due to individual variations and the vascular tissue growth and remodeling. To address this, we propose a novel hybrid computational framework that integrates uncertainty quantification with biomechanical modeling. Our methodology combines three components: (1) an efficient healing model to simulate aneurysm progression, (2) a Legendre polynomial surrogate model and a back-propagation artificial neural network to construct a reduced-order stochastic solver, and (3) the Smolyak algorithm to optimize parameter sampling. This approach achieves a 90% reduction in computational time compared to Monte Carlo methods while maintaining accuracy.