Project Summary Clinical decisions based on hemodynamic data –time-varying measurements of pressure and volume in the cardiovascular (CV) system– can be some of the most complex physicians will face. Yet they are consigned to tackle these decisions with little more than a mental model of the patient. From deceptively simple questions like whether a patient will respond to fluid resuscitation, to plainly challenging ones like whether the right ventricle will tolerate placement of a left ventricular assist device, the physician must mentally reconstruct a patient’s physiology from the shadows it casts in the form of clinical state measurements. Right ventricular (RV) dysfunction, in particular, represents a large class of problems in cardiology whose management is widely seen as challenging. This stems from inadequate RV monitoring tools and the challenge of gauging the RV’s interactions with other cardiovascular subsystems. Given detailed hemodynamic measurements a physician should be able to tap computational resources that would systematically infer a quantitative cardiovascular (CV) model that gave rise to it, thereby characterizing the RV and its interactions. Computational models such as these, which could be repeatedly updated as new sensor data stream in, are known as digital twins. Here we propose to develop and validate a digital “hemodynamic” twin to aid the management of right ventricular failure. Such a computational tool would allow the physician to explore a patient’s right ventricular physiology in silico and to predict the effects and the limits of drug therapy. Specifically, we aim to solve the inverse problem of system identification (SID) on real-world clinical data, both extending the computational tools for hemodynamic SID and validating them on a large number of patients. We will focus on patients who are susceptible to RV dysfunction, including outpatients with pulmonary hypertension and inpatients recovering from cardiac surgery in an intensive care unit. The creation of a digital twin will require only data that are collected using routine hemodynamic monitoring techniques that are widely available. We will test the fidelity of our SID tools, use the digital twins to search for new subtypes of pulmonary hypertension and post-cardiac surgery recovery, and determine the ability of newly identified cardiovascular parameters to predict patient outcomes.