A Digital Twin to Predict Patient Response to Valve Replacement Following Aortic Stenosis

Lack of available data, ease of clinical use and lack of evidence for prognostic benefit are arguably the key limitations to clinical uptake for any model. This paper reports the development of an image-based analysis protocol, coupling 0D left heart and systemic circulation components with a 3D aortic valve model to measure and to predict the pressure gradient across the aortic valve at rest. The model is personalized using routine clinical data available for aortic valve patients, augmented by additional image data (transesophageal echo and/or CT) to support valve characterization. Computed aortic pressure gradient both pre- and post-intervention was compared with clinical measurements based on Doppler ultrasound for a cohort of 21 patients with aortic valve disease. Correlation for the diseased state measures were adequate (R2= 0.81) for those cases for which associated image data was deemed to be of acceptable quality for segmentation to support a 3D computational fluid dynamics analysis but poor otherwise. Post treatment correlation was reasonable (R2= 0.47) for all cases. Importantly, the personalized model presented here describes the interaction between the patient’s cardiovascular system, including the heart and circulation, and the valve, rather than evaluating the valve in isolation.