In vivo evaluation of two adaptive Starling-like control algorithms for left ventricular assist devices

The implantation of a left ventricular assist device (LVAD) is often the only therapy in terminal heart failure (HF). However, despite technical advancements, the physical fitness of the patients is still limited. One strategy to improve the benefits of ventricular assist device therapy might be the implementation of load adaptive control strategies. Two control strategies and a constant speed controller (CS) were implemented in an acute animal model where four healthy pigs received LVAD implantations. In the first strategy (preload recruitable stroke work [SW] controller, PRS), the desired pump work was computed in relation to the end-diastolic volume. In the second strategy, the controller was programmed to keep a fixed ratio of the mean hydraulic power of the assist device to the mean hydraulic power of the left ventricle (power relation controller, PR). Preload reduction, afterload increase experiments and short-term coronary artery occlusions were conducted to test the behavior of the control strategies under variable conditions. Within the experiments, the PR controller demonstrated the best preload sensitivity. The PRS controller had the best response to an increased afterload and to a reduced ventricular contractility in terms of effectively preventing ventricular overloading and increasing VAD support. No significant differences in systemic flow were observed.