Validation of a Non-invasive Inverse Problem-Solving Method for Stroke Volume

• Published in: Frontiers in physiology 2022-01, Vol.12, p.798510-798510
• Main Authors: Bikia, Vasiliki, McEniery, Carmel M, Roussel, Emma Marie, Rovas, Georgios, Pagoulatou, Stamatia, Wilkinson, Ian B, Stergiopulos, Nikolaos
• Format: Article
• Language: English
• Subjects: cardiac output; data assimilation; mathematical modeling; non-invasive monitoring; Physiology; vascular aging
• ISSN: 1664-042X; 1664-042X
• DOI: 10.3389/fphys.2021.798510

Stroke volume (SV) is a major biomarker of cardiac function, reflecting ventricular-vascular coupling. Despite this, hemodynamic monitoring and management seldomly includes assessments of SV and remains predominantly guided by brachial cuff blood pressure (BP). Recently, we proposed a mathematical inverse-problem solving method for acquiring non-invasive estimates of mean aortic flow and SV using age, weight, height and measurements of brachial BP and carotid-femoral pulse wave velocity (cfPWV). This approach relies on the adjustment of a validated one-dimensional model of the systemic circulation and applies an optimization process for deriving a quasi-personalized profile of an individual's arterial hemodynamics. Following the promising results of our initial validation, our first aim was to validate our method against measurements of SV derived from magnetic resonance imaging (MRI) in healthy individuals covering a wide range of ages ( = 144; age range 18-85 years). Our second aim was to investigate whether the performance of the inverse problem-solving method for estimating SV is superior to traditional statistical approaches using multilinear regression models. We showed that the inverse method yielded higher agreement between estimated and reference data ( = 0.83, < 0.001) in comparison to the agreement achieved using a traditional regression model ( = 0.74, < 0.001) across a wide range of age decades. Our findings further verify the utility of the inverse method in the clinical setting and highlight the importance of physics-based mathematical modeling in improving predictive tools for hemodynamic monitoring.