A Hybrid Experimental-Computational Modeling Framework for Cardiovascular Device Testing

Abstract

Significant advances in biomedical science often leverage powerful computational and experimental modeling platforms. We present a framework named physiology simulation coupled experiment (“PSCOPE”) that can capitalize on the strengths of both types of platforms in a single hybrid model. PSCOPE uses an iterative method to couple an in vitro mock circuit to a lumped-parameter numerical simulation of physiology, obtaining closed-loop feedback between the two. We first compared the results of Fontan graft obstruction scenarios modeled using both PSCOPE and an established multiscale computational fluid dynamics method; the normalized root-mean-square error values of important physiologic parameters were between 0.1% and 2.1%, confirming the fidelity of the PSCOPE framework. Next, we demonstrate an example application of PSCOPE to model a scenario beyond the current capabilities of multiscale computational methods—the implantation of a Jarvik 2000 blood pump for cavopulmonary support in the single-ventricle circulation; we found that the commercial Jarvik 2000 controller can be modified to produce a suitable rotor speed for augmenting cardiac output by approximately 20% while maintaining blood pressures within safe ranges. The unified modeling framework enables a testing environment which simultaneously operates a medical device and performs computational simulations of the resulting physiology, providing a tool for physically testing medical devices with simulated physiologic feedback.

Risks and Benefits of Using a Commercially Available Ventricular Assist Device for Failing Fontan Cavopulmonary Support: A Modeling Investigation

“…Currently, the cavopulmonary device is not employed in the clinical management of Fontan patients, therefore the results of this study cannot be validated against clinical measurements. Fluctuation in the rotational speed of the pump as a result of pulsatile flow has been observed in our previous experimental study [33]; while the LPN model is not capable of capturing these fluctuations, since the pulsatility in the pulmonary flow is low, these fluctuations should not result in significant error in the simulated hemodynamic outcomes. Another limitation of the current study is that the LPN model is not able to resolve vessel wall shear stress, which is known to impact endothelial function and pulmonary vascular resistance in Fontan patients.…”

An algorithm for coupling multibranch in vitro experiment to numerical physiology simulation for a hybrid cardiovascular model

“…The performance curves of HeartWare describing the pressure increase produced by the VAD ( ∆P ) based on the flow ( Q ) passing through it at various rotation speeds are obtained from Reference . A , B and C are coefficients determined experimentally during pump characterization . The pressure drop across a stenosis can be modeled as a quadratic function of flow .…”

“…19,27 The pressure drop across a stenosis can be modeled as a quadratic function of flow. 9,19 We used a pressure drop coefficient of k = 0.0004 mm Hg s 2 ml −2 to emulate realistic pressure drops across a vascular stenosis. Linear resistances are utilized to model pressure loss through blood vessels, which is linearly proportional to the flow rate (Q).…”

“…To this end, we have previously introduced the Physiology Simulation Coupled Experiment framework and demonstrated the coupling of a 2‐branch (one inlet and one outlet) in vitro experiment to an LPN physiology simulation using a proportional‐control‐based coupling algorithm. In the present study, we introduce a Broyden‐method‐based algorithm capable of integrating a multibranch (ie, more than two branches) experiment with a computational physiology simulation, enabling the closed‐loop coupling of the two.…”

Hardware in the Loop Implementation of the Oscillator-based Heart Model: A Framework for Testing Medical Devices

“…These methodologies, born mainly for the aerospace and power electronics sector, are also extending to other electronic devices, especially for critical applications. Think for example of medical therapy devices that must interact with a person or part of a person [6][7][8]. During the development and validation phase of medical devices, the patient cannot be involved in the design process for several reasons including the fact that the pathology to be resolved must be induced in the patient just when the device is being tested, which is not easy to implement.…”