["The Journal of Physiology, EarlyView. ", "\nAbstract figure legend A cohort of five patients diagnosed with aortic stenosis (AS) was considered. Computed tomography (CT) scan data and 4D flow magnetic resonance imaging (MRI) are available for these patients. CT data are utilised to segment and reconstruct the aorta's anatomy and the aortic valve (AV) leaflets. A region of interest (ROI) is defined past the AV, and the flow extracted in various planes within the ROI is averaged. This average flow is subsequently employed as the inlet boundary condition for the aorta in the patient‐specific fluid‐structure interaction (FSI) model. The model has been calibrated for each patient: distinct values of the Young modulus of the AV have been selected, and the optimal value corresponds to the one resulting in the most accurate average velocity within the ROI (comparing FSI with 4D flow MRI). Once calibrated, the model has been validated in terms of various quantities of interest (QoIs), including velocity, vorticity, mean kinetic energy, wall shear stress, scalar shear stress, enstrophy, and aortic valve area. Once validated, the model is utilised to extract additional QoIs pertinent to the context of AS assessment, such as blood residence time and transvalvular pressure gradient.\n\n\n\n\nAbstract\nSystematic in vivo validations of computational models of the aortic valve (AV) remain scarce, despite successful validation against in vitro data. Utilizing a combination of computed tomography and 4D flow magnetic resonance imaging data, we developed patient‐specific fluid–structure interaction models of the AV immersed in the aorta for five patients in the pre‐transcatheter AV replacement configuration. Our computational models are subjected to rigorous validation against 4D flow measurements. Our results demonstrate the models’ capacity to accurately replicate flow dynamics. In addition, we illustrate how computational models can serve as valuable cross‐checks to reduce noise and erratic behaviour of in vivo data. Crucially, our validated models enable the measurement of additional critical quantities essential for a comprehensive understanding of aortic stenosis (AS) and its treatments: we compute the blood residence time, enhancing precision and personalization in assessing the probability of thrombus formation within the aorta. This study represents a significant step towards integrating in silico technologies into real clinical contexts, providing a robust framework for improving AS diagnosis and the design of next‐generation AV bioprostheses.\n\n\n\n\n\n\n\n\n\nKey points\n\nPatient‐specific fluid–structure interaction computational models of the aortic valve are developed for five patients in pre‐transcatheter aortic valve replacement configuration.\nA synergistic approach involving in silico models and in vivo data is utilized, including computed tomography and 4D flow magnetic resonance imaging.\nA patient‐specific calibration strategy is introduced to identify the aortic valve Young's modulus, leveraging in vivo flow‐derived metrics in combination with patient‐specific valve and aortic geometries.\nThe computational models are validated against in vivo 4D flow measurements, demonstrating their ability to replicate flow dynamics accurately.\nThe potential of computational models to cross‐check and reduce noise in in vivo data is highlighted, providing additional critical physiological quantities for comprehensive aortic stenosis assessment, such as blood residence time, important for thrombus formation evaluation.\n\n\n"]