SCH: Simulation Optimization of Cardiac Surgical Planning

Many patients take surgical interventions to fight the battle against heart disease. Surgical successes are critical to the patients’ health and their family well-being. For e.g., atrial fibrillation (AF) is the most common arrhythmia in elder population. Catheter ablation is an established treatment for AF, which sequentially creates incision lines to block faulty electrical pathways. However, there are large variations in surgical outcomes. Modern healthcare systems are investing heavily in sensing and computing technology to increase information visibility and cope with disease complexity. Massive data are readily available in the surgical environment. Realizing the full data potential for optimal decision support depends on the advancement of information processing and computational modeling methodologies. Our long-term goal is to advance the frontier of precision cardiology by developing new sensor-based modeling and simulation optimization methodologies. The objective of this project is to optimize AF ablation by integrating simulation-enabled planning with physics-augmented machine learning of sensor signals from patients who underwent AF ablation. This objective will be accomplished by pursuing 3 specific aims: 1) Physics-augmented artificial intelligence (AI) for cardiac modeling – This approach will assimilate heterogeneous sensing data and incorporate electrophysiology prior knowledge into deep learning to increase the robustness of decision making under uncertainty, thereby driving computer simulation into clinical applications; 2) Optimal sensing and sequential learning of space-time AF dynamics – This approach will provide quantitative knowledge of disease mechanisms instead of subjective knowledge that is difficult to translate (or transfer), thereby reducing healthcare disparity due to the availability of human experts in rural areas; 3) Integrating sensor-based learning and simulation optimization for surgical planning - This approach will integrate physics-augmented modeling (Aim 1) and sensor-based learning (Aim 2) with simulation optimization to improve the clinical practice towards data-driven & simulation-guided surgical planning. This project will make a major breakthrough towards precision cardiology by (i) going beyond the current practice of largely expert-based or ad hoc decisions, (ii) capturing underlying complexities in space-time cardiac dynamics, and (iii) integrating physics-based modeling, sensor-based learning, and simulation-based planning for surgical decision support.