Abstract
A continuous, real-time monitoring method for the cardiac output during assistance with left ventricular assist device (LVAD) was developed based on time series analysis. During ejection of LVAD, parametric model for arterial load was identified by least square algorithm. Natural heart flow was estimated in its ejection period with use of the identified model inversely assuming the estimated parameters to be invariable during the cardiac cycle. Classical windkessel model with some modification was used for the estimation. In animal experiments cardiac output were changed by abrupt loading and unloading with 500ml of blood. Natural heart flow were well estimated with the counter pulsation with the delay time of 50%. However, the waveforms showed oscillation compared with measured ones. It is concluded that stroke volume of natural heart and cardiac output could be predicted well by this method.
