Model-based prediction of a percutaneous ventricular assist device performance

A percutaneous ventricular assist system is an external heart assist device that bypasses blood from the left atrium and returns it to the femoral artery to support patients who suffer from acute heart failure. The system consists of a centrifugal blood pump, an atrial drainage cannula, and various sizes of arterial perfusion cannula. Because the device allows cardiologists the freedom choosing the arterial cannula based on a patient's body size, it is extremely difficult but important to predict the level of support the device can provide to the patient before the devise is up and running. In this paper, the TandemHeart pVAD (Cardiacassist Inc. Pittsburgh, PA) is modeled as a nonlinear electric circuit, including a speed dependent voltage source and current dependent resistors to predict the performance of the system by specifying pump speed, mean arterial pressure (MAP), and mean left atrial pressure (LAP). The model structure is developed based upon the pipeline theory while the model parameters are identified by least-squares fit of the model to the experimental data. The flow rate is predicted by solving a quadratic equation while coefficients in the equation are scaled, depending on the arterial cannula configurations. The model can predict the flow rates accurately with error indices of all test conditions less than 6%, comparing the predicted flow from the model with the experimental data.