Primary mitral valve regurgitation (MR) is a common heart disease that induces volume
overload (VO) and cardiac remodeling. The most effective treatment for MR is mitral valve
repair (MVr). However, clinical outcomes of MVr are often suboptimal, even in asymptomatic
patients with functional hearts, 20% of whom may experience systolic dysfunction within 12
months. Recent research has highlighted various clinical markers that correlate with poor MVr
outcomes but cannot reliably predict long-term response for individual patients. Understanding how neurohormonal factors, mechanics, and remodeling determine MVr outcomes may require more complex, integrative approaches than currently exist. So, I propose enhancing multiscale computational modeling with machine learning to 1) predict changes in LV size and function following MVr and 2) identify pathways associated with remodeling and deterioration of cardiac function in the context of MVr.  As a first step, we have developed a multiscale model of cardiac function and remodeling encompassing ventricular mechanics, circulation hemodynamics, and a network model of cardiomyocyte molecular signaling pathways. Then, we used a Markov Chain Monte Carlo algorithm to integrate data from 76 experimental studies of volume overload. The resulting calibrated model can generate accurate probabilistic predictions on the effect of drug and hormonal interventions within and outside the context of experimental volume overload and can predict the reversal of cardiac hypertrophy following mitral valve replacement in dogs. Furthermore, this model identifies a molecular pathway that could link the loss of ventricular contractile function to exacerbated ventricular hypertrophy and impaired recovery following mitral valve repair.