Software Tool for Optimization of CRT

ABSTRACT Heart failure (HF) is a nationwide epidemic. Many HF patients are presented with left bundle branch block (LBBB) morphology and mechanical dyssynchrony, which are strong predictors of cardiovascular mortality. For patients with prolonged QRS duration >150 ms, cardiac resynchronization therapy (CRT) is recommended to restore normal activation patterns in the heart. About 30% of patients, however, still do not improve after CRT (non-responders), and the percentage of CRT non-responders has remained relatively constant. To address the issue of CRT non-responders and improve the treatment of conduction disorders, there are new exciting developments in pacing therapy (e.g., HIS bundle pacing, left branch bundle (LBB) pacing, LV endocardial pacing). When ischemia and/or a scar are present, experiments have shown that lead positioning and timing of the LV stimulation are critical parameters that require precise optimization to achieve positive CRT responses. This is especially relevant when a large number of CRT recipients (≥50%) in North America and Europe have ischemic cardiomyopathy (ICM), and the percentage of CRT non-responders is higher in ICM patients than in non-ICM patients. As such, computational models are powerful tools that can be exploited to optimize CRT response on a patient-specific basis. In this Phase I grant, we will develop a software tool for the optimization of therapies to reduce non-responder rates and patient identification for CRT, conduction system pacing (CSP), and LV endocardial pacing (LVEP) therapies using physics-based modeling, artificial intelligence, and machine learning (ML). Coronary flow and electromechanics models will be coupled bi-directionally. We pioneered modeling of this coupling that includes coronary flow regulation mechanisms. These studies suggest a threshold in coronary perfusion, depending on the ischemia location and size, where CRT is ineffective (i.e., “point of no return”). With patient-specific data, we can translate our swine-specific models to the clinic, as we did for our study on the effects of an LV assist device on a failing human biventricular unit. This work will be accomplished by the following two Specific Aims: (1) Modify an existing experimentally-validated cardiac electro-mechanics-perfusion computational framework to simulate the acute effects of CRT, CSP, and LVEP in treating mechanical dyssynchrony in LBBB + ischemia; and (2) Integrate the computational modeling framework with ML and optimization algorithms to optimize CRT, CSP, and LVEP in patients with ischemia. This proposal will provide a software tool for optimizing therapies to reduce non-responder rates and patient identification for CRT, CSP, and LVEP therapies. This would have a substantial impact on improving treatment and reducing the healthcare cost of the HF epidemic. A successful Phase I will pave the way to Phase II to advance this technology through regulatory approval of software as a medical device to improve CRT outcomes.