Abstract
The left ventricular assist device(LVAD) has not only a role as the circulatory support but also as left ventricular unloading.Although high assist rate is effective for ventricular unloading,it may cause aortic flow stagnation,which may cause thrombosis or fusion of the aortic valve.Therefore,it is important to adjust a pump output corresponding to a condition of aortic valve opening.In this study,a method for detecting whether aortic valve is open has been proposed and evaluated.Support Vector Machine(SVM) of which feature variables were a rotational speed and current consumption was used for state classification.In addition,the empirical verification algorithm was applied to SVM output in order to improve sensitivity and specificity.The training and evaluation data were obtained from animal experiments using adult goats.The SVM gave a good performance when second derivative of the variables were used as features.The result indicated that the proposed algorithm have practical performance for aortic valve state detection.The information of an aortic valve state may contribute to development for automatic controller for LVAD.In future study,it is necessarily to investigate an individual difference of the parameters.
