Reduced order digital twins of cardiovascular devices and complex congenital diseases

Almost 1% of neonates are born each year with clinically significant congenital heart disease (CHD), making it the most prevalent birth defect worldwide. Among these, single ventricle (SV) malformations and severe right heart obstructions are among the most challenging, often requiring complex, multi-stage surgical reconstruction or long-term mechanical circulatory support (MCS). The management of such conditions is complicated by patient-specific anatomical variability and the lack of large clinical datasets, which limits the ability to generalize treatment protocols. In recent years, physics-based computational modeling has emerged as a promising tool to support personalized decision-making in both surgical planning and device design. This thesis explores such approaches, combining cardiovascular simulations, virtual patient modeling, and mechanical support optimization to address key clinical challenges in the treatment of CHD. The work is divided into two main parts. The first part focuses on pre-operative planning for patients with pulmonary atresia and an intact ventricular septum—a rare condition characterized by a severely underdeveloped right ventricle (RV). In these cases, choosing between biventricular repair and one-and-a-half ventricle palliation is often uncertain. To assist in this decision-making, a lumped parameter cardiovascular model was developed and used to simulate virtual patient populations with varying anatomical and physiological characteristics. For each patient, both surgical strategies were evaluated computationally, assessing post-operative pressure, flow, and oxygen delivery across the circulatory system. The results revealed patterns linking RV size and function with surgical success, supporting the potential of in silico models to act as screening tools prior to clinical intervention. This personalized modeling framework can be particularly valuable in rare congenital diseases, where evidence-based guidelines are limited. The second part of the thesis shifts focus to MCS strategies for both neonates and older children with SV physiology, particularly those at risk of surgical failure or in need of transplantation. First, a modified Norwood circulation was investigated in which pulmonary blood flow is mechanically assisted using a circulatory pump. Computational simulations showed that this approach could improve systemic perfusion and oxygenation, especially in patients with compromised ventricular function, offering an alternative pathway for high-risk neonates. Building on this, a novel support concept was proposed for patients with failing Fontan circulation, where the native SV is reassigned to support venous return while a mechanical assist device maintains systemic output. This configuration was tested in various failure scenarios using a validated cardiovascular model and showed promise as a more physiological and efficient alternative to traditional dual-pump systems. Finally, the thesis presents a new framework for designing and optimizing ventricular assist devices by integrating cardiovascular modeling with pump design software. This approach allows for efficient evaluation of pump performance under realistic physiological conditions and supports the development of patient-specific device configurations that balance flow quality with safety and efficiency. Altogether, this thesis demonstrates how integrated computational modeling can support both the clinical and engineering aspects of congenital heart disease management. By creating virtual patient populations, simulating surgical and device outcomes, and optimizing circulatory configurations, the work provides a pathway toward more personalized, predictive, and effective treatment strategies for some of the most complex conditions in pediatric cardiology.