Grading the severity of diastolic dysfunction using echocardiography remains challenging, requiring a complex multiparametric algorithm and often yields inconclusive results [1]. The hallmark of diastolic dysfunction is impaired ventricular relaxation and increased myocardial stiffness, which elevate the left ventricular (LV) filling pressures [2]. Elevated LV end-diastolic pressure (LVEDP) is a clinically recognised marker of LV filling pressure, and LVEDP is directly measured by cardiac catheterisation, which is invasive, resource-intensive, and not suitable for routine screening. Therefore, a non-invasive method to estimate LVEDP is an unmet clinical need for diagnosing diastolic dysfunction. Recent advances in high-frame-rate ultrasound imaging have enabled cardiac shear wave elastography (SWE) to measure the speed of shear waves induced by mitral valve closure, propagating along the intraventricular septum, which is proportional to myocardial stiffness. In this study, a physiology-informed closed-loop lumped-parameter circulation model (LPM) establishes the crucial link between the structural measurements obtained from echocardiography and the material assessments of the myocardium obtained using SWE. Model parameters were estimated using a two-stage optimisation strategy: differential evolution in the first stage, followed by a quasi-Newton method to ensure robust convergence. Multi-modal clinical data were collected from 68 patients who underwent cardiac catheterisation, including blood pressure, echocardiography and SWE. Invasive catheter-based LVEDP served as the reference ground truth. LPM-derived LVEDP showed a strong correlation with invasive measurements (r = 0.74, p < 0.001), with no statistically or clinically significant difference between group means (p = 0.30). Diagnostic performance was evaluated using receiver operating characteristic analysis, yielding an area under the curve (AUC) of 0.95, with a sensitivity of 92% and a specificity of 81% for classifying elevated LVEDP. Integrating multimodal clinical data with mechanistic cardiovascular modelling provides an explainable, clinically relevant, non-invasive framework to support the diagnosis and grading of diastolic dysfunction.