Extensive efforts have focused on developing and integrating core cardiac models, particularly in ventricular electrophysiology. However, as there is a notable lack of atrial models in the literature, our research emphasizes electrophysiological atrial models for both normal and pathological states. While collagen fiber-related anisotropic properties have been extensively studied in ventricles, we have refined active and passive mechanical contraction models for better accuracy.
Novel models for active force generation, including reduced force models through Artificial Neural Networks, have been introduced to balance computational efficiency with cellular intricacy. This work extends to active tension models for both atria and ventricles, essential for cardiac electromechanical simulations. Moreover, our investigation has delved into poroelastic and perfusion models for ventricles, shedding light on oxygen and nutrient supply via coronary arteries. We have also formulated comprehensive models for cardiac valve dynamics, covering leaflet movement within blood through 0D and differential models.
Our research revolves around merging electrophysiology and mechanics models, spanning various integration schemes—monolithic, partitioned, and staggered—encompassing ventricles, atria, and the entire heart. A remarkable feat is the creation of a comprehensive electromechanical heart model, representing crucial biomarkers. This progression ranges from basic electromechanical synchronization to intricate fluid-structure interaction and fluid-perfusion coupling.
To surmount algebraic challenges in space-time discretization, we have designed and implemented preconditioners. Solving the electromechanical model involved devising intergrid transfer operators, addressing electrophysiology and mechanics across distinct Finite Element meshes, capturing multiscale complexities. This extends to torso-heart, poroelastic, and perfusion models integrated with the iHEART simulator.
In addressing variability and uncertainty, we have enhanced reduced order models and uncertainty quantification methods using machine learning. This approach facilitates swift assessment of cardiac biomarkers by adjusting physiological parameters. Scientific Machine Learning has been employed for electrophysiology and electromechanics in physiological and pathological scenarios.
In pursuing personalized treatment, we have integrated patient-specific elements such as anatomies and functional data through refined computational pipelines. These pipelines automate simulations from image capture to mesh-based domain generation. Scientific Machine Learning techniques have been advanced, integrating electrophysiological data to personalize models for arrhythmic disorders.
Through extensive clinical collaborations, we have examined various heart pathologies including electrical dysfunctions (ischemia, tachycardia, atrial fibrillation), valve pathologies (aortic and mitral), and coronary occlusions. Our studies encompass numerical analyses and broader impacts on cardiac functionality, mechanics, and fluid dynamics. Innovative solutions have been proposed for compromised ventricular tissue perfusion.
Our project had a substantial impact, spanning academia, healthcare, and public awareness. Initiatives have been launched to promote Mathematics and Science, highlighting the influential role of the iHEART simulator in society's well-being.
