Sensorless Physiologic Control, Suction Prevention, and Flow Balancing Algorithm for Rotary Biventricular Assist Devices

Objective: Rotary biventricular assist devices (BiVAD) are mechanical pumps that are implanted in the left and right ventricles of biventricular failure patients to pump blood and provide mechanical circulatory support. The objective of this paper was to develop and test a novel sensorless control algorithm that simultaneously satisfies the objectives of providing physiologic control (BiVAD flows meet cardiac demand), preventing ventricular suction, and providing balanced left–right (systemic and pulmonary) flows without the use of implantable flow or pressure sensors in the nonlinear, time varying, and discontinuous circulatory system. Methods: The control algorithm consists of two gain-scheduled proportional–integral controllers for left and right ventricular assist devices and only requires intrinsic pump parameters (speed and power) to maintain differential pump speeds ( $\Delta $ RPM $_{L}$ and $\Delta $ RPM $_{R}$ ) above user-defined thresholds to prevent ventricular suction, and average reference pressure heads ( $\Delta \text{P}_{L}$ , $\Delta \text{P}_{R}$ ) to provide physiologic perfusion and balance left–right-sided flow rates. A model-based approach with extended Kalman and Golay–Savitzky filters was used to estimate $\Delta \text{P}_{L}$ and $\Delta \text{P}_{R}$ . Efficacy and robustness of the algorithm were evaluated in silico during simulated rest and exercise test conditions for: 1) excessive $\Delta \text{P}_{L }$ and/or $\Delta \text{P}_{R}$ setpoints; 2) rapid threefold increase in pulmonary vascular or vena caval resistances; 3) transitions from exercise to rest; and 4) ventricular fibrillation. Results and Conclusion: The proposed sensorless BiVAD algorithm successfully prevented suction, restored physiologic perfusion, and inherently maintained left–right-sided balance for all test conditions. Significance: The proposed algorithm does not require any device modification and may be integrated into current clinical BiVADs.