Credibility Assessment of the Patient-Specific Modeling of the Aneurysmal Ascending Thoracic Aorta: Verification, Validation and Uncertainty Quantification.

Abstract

Purpose: Computational modeling holds promise in predicting patient-specific outcomes and guiding clinical decision-making. The patient-specific model forming the basis of a digital twin can be considered biomedical software, thereby necessitating trust in its predictive accuracy.

Methods: This study applies the ASME V&V40 framework to demonstrate the credibility of patient-specific models of aneurysmal thoracic ascending aorta (ATAA) biomechanics. A comprehensive verification, validation, and uncertainty quantification process was performed to evaluate the accuracy of the patient-specific ATAA model.

Results: After implementing the ASME V&V40 standard, the verification errors on the model inputs (i.e., material parameters and hemodynamic variables) resulted in relative errors (RE) < 1%. Validation and its uncertainty quantification of the output aneurysm diameter response showed area metric errors below 5% in the majority of cases, highlighting the accuracy of the patient-specific ATAA model against the clinical comparator. Uncertainties in wall stress predictions due to model inputs were also quantified by probability density functions. Sensitivity analysis revealed that the unknown value of aneurysm wall thickness drives the model output at the highest extent.

Conclusions: These findings contribute to a standardized methodology for evaluating the credibility of patient-specific models, enhancing their utility in computer-based clinical decision support systems for managing patients with ATAAs.