TwinHeart

Heart failure (HF) remains a leading cause of morbidity and mortality globally, with treatments like ventricular assist devices (VADs) often causing complications such as thrombosis, hemolysis, and end-organ dysfunctions. Addressing these challenges by improving VAD performance and patient outcomes is critical for advancing heart failure management and reducing the societal burden.

This project introduces a novel treatment strategy for HF based on intra-aortic rotary axial pumps, optimized through a high-fidelity computational framework that integrates fluid-structure interaction (FSI) simulations, advanced blood rheology modeling, and multiscale analysis to account for the complex interplay between the pump, arterial tissue, and blood flow. By considering factors such as shear stress, vascular compliance, and blood damage, this approach provides a highly accurate representation of real-world conditions, a significant improvement over traditional idealized models.

Deep reinforcement learning (DRL) will be employed to optimize pump parameters in real time, ensuring personalized adjustments to cardiac function and minimizing risks such as blood damage and end-organ dysfunction. Moreover, graph neural networks (GNNs) trained with CFD data will allow for fast predictions of flow behavior and pump performance, addressing computational constraints and enabling a wider exploration of design configurations.

The project will also include experimental validation through a 3D-printed aortic model, capturing the non-Newtonian properties of blood flow and turbulence transition in both steady and pulsatile flows.

For the first time, this approach combines cutting-edge machine learning, advanced simulation, and experimental techniques to develop a digital twin for patient-specific VAD solutions. The framework not only improves predictions of blood flow dynamics and device performance but also reduces surgical invasiveness and long-term complications. This research has the potential to reshape heart failure treatment by offering highly personalized, optimized care, reducing healthcare costs, and significantly improving patient quality of life and outcomes worldwide.