Model-based assessment of cardiovascular health from noninvasive measurements

Cardiovascular health is currently assessed through a variety of hemodynamic parameters, many of which can only be determined by invasive measurement often requiring hospitalization. A noninvasive method of evaluating several of these parameters such as systemic vascular resistance (SVR), maximum left ventricular elasticity (E(LV)), end diastolic volume (VED), and cardiac output, is presented. The method has three elements: (1) a distributed model of the human cardiovascular system (Ozawa et aL, Ann. Biomed. Eng. 29:284-297, 2001) to generate a solution library that spans the anticipated range of parameter values, (2) a method for establishing the multidimensional relationship between features computed from the arterial blood pressure and/or flow traces (e.g., mean arterial pressure, pulse amplitude, mean flow velocity) and the critical hemodynamic parameters, and (3) a parameter estimation method that yields the best fit between measured and computed data. Sensitivity analyses were used to determine the critical parameters, and the influence of fixed model parameters. Using computer-generated brachial pressure and velocity profiles (which can be measured noninvasively), the error associated with this method was found to be less than 3% for SVR, and less than 10% for E(LV) and V(ED). Simulations were also performed to test the ability of the approach to predict changes in SVR and E(LV) from an initial base line state.