HeartCyPert: Design and Analysis of a Heart Cyber Expert System
By Ehsan Esmaili and Talal Moukabary
Chronic heart failure (CHF) is a leading cause of morbidity and mortality worldwide with over 600,000 new cases diagnosed annually in the United States alone, and the most common cause of sudden death in CHF patients are ventricular arrhythmias. We do not know how to predict which patients with CHF will die from sudden death except that they all have poor heart or left ventricular function.
The goal of this project is to develop a 3D computational model to predict if patients are at high-risk of a ventricular arrhythmia. Currently, developing a 3D model that characterizes the electrical activity in the heart is an extremely computationally challenging task; therefore, cardiac researchers have chosen to use less accurate models that are computationally tractable. In this project, we are developing a data-driven engine that uses data from animals and patients with CHF to gather parameters for a 3D cardiac model that can be used to predict if the patient is at high-risk of a ventricular arrhythmia, which will aid researchers in developing a personalized treatment plan.
The ultimate goal of our research is to develop a cognitive software environment, a Heart Cyber Expert (HeartCyPert, see figure), that can use IBM Watson-like cognitive computing, and data mining and search algorithms to help cardiac researchers, clinicians, and device industries develop new approaches to treat patients with CHF, and ultimately provide inference about whether a subject is of high risk of sudden cardiac death. HeartCyPert achieves this goal by using the clinical data collected from rats and pigs, beyond that of ejection fraction, to parameterized a personalized 3D electrophysiological model of the heart. The personalized 3D model can then be used to simulate new effects of new treatments on a subject in silico to verify if the treatment is safe. The development of such a data analytics engine that provides a feedback loop for performing cardiac research with existing clinical data and mathematical models will significantly improve the time and resources of clinicians studying VT, and provide an accurate approach to correctly identifying subjects that are at high risk VT/VF.