Peter J. Ayre

Response of rotary blood pumps to changes in preload and afterload at a fixed speed setting are unphysiological when compared with the natural heart.

Responses of four rotary blood pumps (Incor, Heartmate II, Heartware, and Duraheart) at a single speed setting to changes in preload and afterload were assessed using the human left ventricle as a benchmark for comparison. Data for the rotary pumps were derived from pressure flow relations reported in the literature while the natural heart was characterized by the Frank-Starling curve adjusted to fit outputs at different afterloads reported in the literature. Preload sensitivity (mean ± SD) for all pumps at all afterloads tested was 0.105 ± 0.092 L/min/mm Hg, while afterload sensitivity was 0.09 ± 0.034 L/min/mm Hg-values that were not significantly different (t-test, P = 0.56). By contrast, preload sensitivity of the natural heart was over twice as high (0.213 ± 0.03 L/min/mm Hg) and afterload sensitivity about one-third (0.03 ± 0.01 L/min/mm Hg) the values recorded for rotary pumps (t-test, P < 0.001). Maximum preload sensitivity and minimum afterload sensitivity allow the right and left ventricles to synchronize outputs without neural or humoral intervention. This theoretical study reinforces the need to provide preload sensitive control mechanisms of sufficient power to enable the pump and left ventricle in combination to adapt to changes in right ventricular output automatically.

Non-invasive flow estimation in an implantable rotary blood pump: a study considering non-pulsatile and pulsatile flows

Non-invasive estimation of flow was investigated in an implantable rotary blood pump (iRBP) with a hydrodynamic bearing. The effects of non-pulsatile and pulsatile flows were studied using in vitro mock loops, and acute (N = 3) and chronic (N = 6) ovine experiments. Using the non-pulsatile and pulsatile mock loops an average flow estimation algorithm was derived from root mean square (RMS) pump impeller speed and RMS input power. These algorithms were programmed into the iRBP controller for subsequent validation in vivo. In the acute experiments, venous return and systemic vascular resistance were adjusted through pharmacological intervention and exsanguination to produce an average range of pump flows from 0.0 to 2.6 l min(-1). Over this range the RMS estimation error was 88 +/- 12 ml, with a linear correlation slope of 0.992 +/- 0.006 (R2 = 0.986 +/- 0.004). In the chronic experiments, animals were monitored daily for up to three months and an average range of flows from 2.8 to 4.8 l min(-1) recorded. A linear correlation between the estimated and measured pump flows yielded a slope of 1.005 +/- 0.006 (R2 = 0.966 +/- 0.004). The RMS estimation error was 120 +/- 11 ml. Using this algorithm it is possible to effectively estimate flow in a rotary blood pump without implanting additional invasive sensors.

Effect of alteration in pump speed on pump output and left ventricular filling with continuous-flow left ventricular assist device

Third-generation continuous-flow left ventricular assist devices (LVAD) provide reduced pulsatility flow. We examined the safe working range for LVAD pump speed and the effect on pump output and cardiac function in 13 stable outpatients with VentrAssist-LVAD (Ventracor Ltd, Australia). Pump speed was decreased from a baseline mean of 2,073 ± 86 revolutions per minute (RPM, with corresponding mean flow of 5.59 ± 1.18 L/min, mean ± standard deviation) to an average low-speed of 1,835 ± 55 RPM (corresponding flow 4.68 ± 0.99 L/min) and up to high-speed of 2,315 ± 66 RPM (corresponding flow 6.30 ± 1.29 L/min). There was a strong linear relationship between alteration in speed and flow rates (r2 = 0.89, p < 0.00001) but marked interpatient variation. Downward titration to preset minimum 1,800 RPM was achieved in 9/13 (69%) and upward titration to the preset maximum 2,400 RPM was achieved in 4/13 (31%). Upward titration was stopped due to ventricular suction or nonsustained ventricular tachycardia (VT) in 4/13 (31%). Ventricular suction or VT (in 4/13) tended to be more common in patients with poor right ventricular (RV) function (p = 0.07). In summary, pump flow is stable within a relatively small speed range and should not be altered without close monitoring due to variation in response between patients, particularly with concomitant RV impairment.

Noninvasive Activity-based Control of an Implantable Rotary Blood Pump: Comparative Software Simulation Study

A control algorithm for an implantable centrifugal rotary blood pump (RBP) based on a noninvasive indicator of the implant recipients activity level has been proposed and evaluated in a software simulation environment. An activity level index (ALI)-derived from a noninvasive estimate of heart rate and the output of a triaxial accelerometer-forms the noninvasive indicator of metabolic energy expenditure. Pump speed is then varied linearly according to the ALI within a defined range. This ALI-based control module operates within a hierarchical multiobjective framework, which imposes several constraints on the operating region, such as minimum flow and minimum speed amplitude thresholds. Three class IV heart failure (HF) cases of varying severity were simulated under rest and exercise conditions, and a comparison with other popular RBP control strategies was performed. Pump flow increases of 2.54, 1.94, and 1.15 L/min were achieved for the three HF cases, from rest to exercise. Compared with constant speed control, this represents a relative flow change of 30.3, 19.8, and -15.4%, respectively. Simulations of the proposed control algorithm exhibited the effective intervention of each constraint, resulting in an improved flow response and the maintenance of a safe operating condition, compared with other control modes.

Numerical Optimization Studies of Cardiovascular-Rotary Blood Pump Interaction

A heart-pump interaction model has been developed based on animal experimental measurements obtained with a rotary blood pump in situ. Five canine experiments were performed to investigate the interaction between the cardiovascular system and the implantable rotary blood pump over a wide range of operating conditions, including variations in cardiac contractility and heart rate, systemic vascular resistance (SVR), and total blood volume (V(total) ). It was observed in our experiments that SVR decreased with increasing mean pump speed under the healthy condition, but was relatively constant during the speed ramp study under reduced cardiac contractility conditions. Furthermore, we also found a significant increase in pulmonary vascular resistance with increasing mean pump speed and decreasing total blood volume, despite a relatively constant SVR. Least squares parameter estimation methods were utilized to fit a subset of model parameters in order to achieve better agreement with the experimental data and to evaluate the robustness and validity of the model under various operating conditions. The fitted model produced reasonable agreement with the experimental measurements, both in terms of mean values and steady-state waveforms. In addition, all the optimized parameters were within physiological limits.

NONINVASIVE DETECTION OF SUCTION IN AN IMPLANTABLE ROTARY BLOOD PUMP USING NEURAL NETWORKS

Granting those heart failure patients who are recipients of an implantable rotary blood pump (iRBP) greater functionality in daily activities is a key long-term strategy currently being pursued by many research groups. A reliable technique for noninvasive detection of the various pumping states, most notably that of ventricular collapse or suction, is an essential component of this strategy. Presented in this study is such a technique, whereby various indicators are derived from the noninvasive pump feedback signals, and a suitable computational methodology developed to classify the pumping states of interest. Clinical telemetry data from ten implant recipients was categorized (with the aid of trans-oesophageal echocardiography) into the normal and suction states. These data are used to develop a pumping state classifier based on an artificial neural network (ANN). Nine indices, derived from the noninvasive impeller speed signal, form the inputs to this ANN classifier. During validation, the resulting ANN classifier achieved a maximum sensitivity of 98.54% (609/618 samples of 5 s in length) and specificity of 99.26% (12,123/12,213 samples) for correct detection of the suction state. The ability to detect the suction state with such a high degree of accuracy provides a critical parameter both for control strategy development, and for clinical care of the implant recipient.

Effect of Parameter Variations on the Hemodynamic Response Under Rotary Blood Pump Assistance

Numerical models, able to simulate the response of the human cardiovascular system (CVS) in the presence of an implantable rotary blood pump (IRBP), have been widely used as a predictive tool to investigate the interaction between the CVS and the IRBP under various operating conditions. The present study investigates the effect of alterations in the model parameter values, that is, cardiac contractility, systemic vascular resistance, and total blood volume on the efficiency of rotary pump assistance, using an optimized dynamic heart-pump interaction model previously developed in our laboratory based on animal experimental measurements obtained from five canines. The effect of mean pump speed and the circulatory perturbations on left and right ventricular pressure volume loops, mean aortic pressure, mean cardiac output, pump assistance ratio, and pump flow pulsatility from both the greyhound experiments and model simulations are demonstrated. Furthermore, the applicability of some of the previously proposed control parameters, that is, pulsatility index (PI), gradient of PI with respect to pump speed, pump differential pressure, and aortic pressure are discussed based on our observations from experimental and simulation results. It was found that previously proposed control strategies were not able to perform well under highly varying circulatory conditions. Among these, control algorithms which rely on the left ventricular filling pressure appear to be the most robust as they emulate the Frank-Starling mechanism of the heart.

Noninvasive Pulsatile Flow Estimation for an Implantable Rotary Blood Pump

A noninvasive approach to the task of pulsatile flow estimation in an implantable rotary blood pump (iRBP) has been proposed. Employing six fluid solutions representing a range of viscosities equivalent to 20-50% blood hematocrit (HCT), pulsatile flow data was acquired from an in vitro mock circulatory loop. The entire operating range of the pump was examined, including flows from -2 to 12 L/min. Taking the pump feedback signals of speed and power, together with the HCT level, as input parameters, several flow estimate models were developed via system identification methods. Three autoregressive with exogenous input (ARX) model structures were evaluated: structures I and II used the input parameters directly; structure II incorporated additional terms for HCT; and the third structure employed as input a non-pulsatile flow estimate equation. Optimal model orders were determined, and the associated models yielded minimum mean flow errors of 5.49% and 0.258 L/min for structure II, and 5.77% and 0.270 L/min for structure III, when validated on unseen data. The models developed in this study present a practical method of accurately estimating iRBP flow in a pulsatile environment.

Classification of Physiologically Significant Pumping States in an Implantable Rotary Blood Pump: Patient Trial Results

An integral component in the development of a control strategy for implantable rotary blood pumps is the task of reliably detecting the occurrence of left ventricular collapse due to overpumping of the native heart. Using the noninvasive pump feedback signal of impeller speed, an approach to distinguish between overpumping (or ventricular collapse) and the normal pumping state has been developed. Noninvasive pump signals from 10 human pump recipients were collected, and the pumping state was categorized as either normal or suction, based on expert opinion aided by transesophageal echocardiographic images. A number of indices derived from the pump speed waveform were incorporated into a classification and regression tree model, which acted as the pumping state classifier. When validating the model on 12,990 segments of unseen data, this methodology yielded a peak sensitivity/specificity for detecting suction of 99.11%/98.76%. After performing a 10-fold cross-validation on all of the available data, a minimum estimated error of 0.53% was achieved. The results presented suggest that techniques for pumping state detection, previously investigated in preliminary in vivo studies, are applicable and sufficient for use in the clinical environment.

Non-invasive estimation and control of inlet pressure in an implantable rotary blood pump for heart failure patients

We propose a dynamical model for mean inlet pressure estimation in an implantable rotary blood pump during the diastolic period. Non-invasive measurements of pump impeller rotational speed (ω), motor power (P), and pulse width modulation signal acquired from the pump controller were used as inputs to the model. The model was validated over a wide range of speed ramp studies, including (i) healthy (C1), variations in (ii) heart contractility (C2); (iii) afterload (C2, C3, C4), and (iv) preload (C5, C6, C7). Linear regression analysis between estimated and extracted mean inlet pressure obtained from in vivo animal data (greyhound dogs, N = 3) resulted in a highly significant correlation coefficients (R(2) = 0.957, 0.961, 0.958, 0.963, 0.940, 0.946, and 0.959) and mean absolute errors of (e = 1.604, 2.688, 3.667, 3.990, 2.791, 3.215, and 3.225 mmHg) during C1, C2, C3, C4, C5, C6, and C7, respectively. The proposed model was also used to design a controller to regulate mean diastolic pump inlet pressure using non-invasively measured ω and P. In the presence of model uncertainty, the controller was able to track and settle to the desired input within a finite number of sampling periods and minimal error (0.92 mmHg). The model developed herein will play a crucial role in developing a robust control system of the pump that detects and thus avoids undesired pumping states by regulating the inlet pressure within a predefined physiologically realistic limit.