P79: Comparison of Cardiovascular Models for Validation of Controlled Ventricular Assist Devices

Abstract

Purpose of the Study: In the evaluation of control schemes for Ventricular Assist Devices (VADs), it is important to utilize a precise model of the human cardiovascular system for both in silico and in vitro testing. As various patients and experimental data result in varying performance of cardiovascular models (CVM) and their respective parameters, it is necessary to assess the robustness of the control schemes through examination with a broad range of parameter sets. To achieve this, we conducted a comparison of two state-of-the-art CVMs and analyzed their interaction with a controlled VAD under dilated heart conditions. Methods: In this study, two state-of-the-art CVMs were investigated: the model developed by Colacino et al. (2007) (COLA) and the model by Lim et al. (2010) (LIM). Both CVMs were implemented as real-time capable models and evaluated both in silico and in vitro using our hybrid mock circulation test bench. Similar test cases were established for both models, encompassing a rest-to-exercise scenario and variations of preload and afterload conditions. These test cases were simulated and experimentally tested, with the results of the left ventricular (LV) pressure, mean aortic (MAP) pressure, cardiac output (CO), pressure-volume (PV) loops facilitating the analysis of the response of the dilated heart supported with a controlled VAD, and the comparison with the response of the healthy heart (HH). We tested six different physiological controllers and the influence of the models on the controller’s reaction. Results: During the rest-to-exercise experiment, the comparison of the two CVMs revealed comparable results in terms of MAP and CO. However, the pressure transition from systole to diastole in the COLA model was found to be faster than in the LIM model. Under the same conditions, the PV loops also exhibited a similar variation, showing differences in shape as well as volume shift. For the rest-to-exercise conditions, the COLA model achieved a higher LV volume compared to the LIM model due to the assumption of different unstressed venous volumes (approx. 20% less for the LIM model). For the rest conditions, the CO was comparable to that of the HH in both models. Conversely, during the exercise condition, the COLA model showed a deficit of approx. 2.2 L/min, while the LIM model was relatively close to the CO of the HH. The systolic pressure controller, as one of the physiological controllers tested, results in CO of about 8.7 L/min (COLA) and about 6.9 L/min (LIM) under exercise conditions, whereas the MAP is nearly the same at about 80 mmHg for both models. Conclusion: The distinct parameters and variations in the structure of the CVMs under investigation resulted in varying blood volumes in the compartments, leading to different pressures during the cardiac cycle and diverse CO values for the dilated heart. This implies differing operating conditions for evaluating the VAD controllers and highlights the need to broaden the test space.Figure 1. Comparison of the COLA (left) and LIM (right) CVM for the rest-to-exercise scenario of a dilated heart with an ejection fraction of around 23 %. The left ventricular pressure (plv), left ventricular volume (Vlv), mean aortic pressure (pMAP), cardiac output of the dilated heart (CO) and healthy heart (COref) are shown.