Application of a search algorithm using stochastic behaviors to autonomous control of a ventricular assist device

A ventricular assist device (VAD) is a device with mechanical pumps implanted adjacent to the patient's native heart to support the blood flow. Mechanical circulatory support using VADs has been an essential therapeutic tool for patients with severe heart failure waiting for a heart transplant in clinical site. Adaptive control of VADs that automatically adjust the pump output with changes in a patient state is one of the important approaches for enhanced therapeutic efficacy, prevention of complications and quality of life improvement. However adaptively controlling a VAD in the realistic situation would be difficult because it is necessary to model the whole including the VAD and the cardiovascular dynamics. To solve this problem, we propose an application of attractor selection algorithm using stochastic behavior to a VAD control system. In this study, we sought to investigate whether this proposed method can be used to adaptively control of a VAD in the simple case of a continuous flow VAD. The flow rate control algorithm was constructed on the basis of a stochastically searching algorithm as one example of application. The validity of the constructed control algorithm was examined in a mock circuit. As a result, in response to a low-flow state with the different causes, the flow rate of the pump reached a target value with self adaptive behavior without designing the detailed control rule based on the experience or the model of the control target.