Key result
Personalized digital twin model predicts pre-intervention aortic pressure gradients with an R² of ~0.81.
Why the study?
Lack of available data, ease of clinical use, and lack of evidence for prognostic benefit are key limitations to the clinical uptake of predictive models.
Can a personalized digital twin model accurately predict the pressure gradient across the aortic valve pre- and post-replacement in patients with severe aortic stenosis?
Can a personalized digital twin model accurately predict the pressure gradient across the aortic valve pre- and post-replacement in patients with severe aortic stenosis?
A personalized digital twin combining 0D circulation and 3D valve models can reasonably predict post-intervention pressure gradients in aortic stenosis, though pre-intervention accuracy heavily depends on image quality.
Image quality limits pre-intervention accuracy; leaves open clinical adoption of digital twin models pending prospective validation.
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.
No takes yet. Share an insight, caveat, or question.
Czechowicz et al. (2026) studied Aortic valve disease (n=21). Digital twin computational model vs. Clinical measurements (Doppler ultrasound) was evaluated on Correlation between computed and clinically measured aortic pressure gradient (R2 = 0.81). The personalized digital twin model adequately predicted pre-intervention aortic pressure gradients (R2 = 0.81) for cases with acceptable image quality, and reasonably predicted post-treatment gradients (R2 = 0.47).
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: