In a joint project between Micromed Technology, Inc we developed an automatic speed adaptation system for rotary blood pump. The Micromed DeBakey VAD®

The system is able to deliver a high pump flow, even in case of an impaired right ventricle and it provides appropriate reaction on ventricular collapse, which may not only damage blood and the intra-ventricular wall, but also cause right heart failure due to septum shift and subsequent tricuspid insufficiency. Further more it is able to adapt to certain changes in fluid balance.

The control system consists of four interacting components. First, a closed loop control circuit with four different set values. A desired pump flow value dependent of the actual heart rate, a desired P2P-Flow value automatically adapted by the suction detection unit and the two speed variation functions and threshold values for maximal pump power and minimal pump flow. Second, a numerically optimized expert system for suction detection. This ensures immediate reduction of pump speed in case of ventricular collapse as well as adaptation of the desired P2P-Flow value in order to avoid repeated suction events. Third, the regular speed variation function 1, which observes the surroundings of the actual working point to determine if corrections are necessary. Simultaneously the efficiency of the actual working point is evaluated and based on these results the desired P2P-Flow value will be adapted. Lastly, a second speed variation function is implemented, which will be activated in case of set value changes based on previous suction detection. Therefore this function observes if additional venous return is available and occasionally decreases the desired P2P-Flow value.

The system was tested in-vitro on a purpose made mock loop and approved by the ethics committee in Vienna for a first clinical trial. Currently 14 patient are included in this first ongoing study and up to know 40 control sessions are done. The test includes application of the control system at 1) Early postoperative phase on ICU

2) Daily life of the patient on regular ward

3) Bicycle ergometry of the recovered patient after rehabilitation

4) Bicycle ergometry during invasive haemodynamic monitoring and spirometric data acquisition

This is an example for successful suction detection at a patient in early postoperative phase on intensive care unit. The suction event was appropriately detected and automatic speed reduction until clearance of suction occurred. What can also be seen in the upper graph is that sever suction can not only reduce pump flow but also can cause arrhythmic beats. In the lower graph the influence of the suction event on the arterial pressure can be seen.

In patients who had a elevated pulmonary resistance before implantation of the LVAD application of a right heart catheter is necessary after recovery to get an approval for heart transplantation. These patients underwent two exercise tests, one in constant speed mode and one in controlled mode with a break of 45 minutes between following this protocol:





1)Preparation for exercise and recording of baseline parameters

2)Warm up: 5 minutes bicycling without load

3)Exercise: 5 minute steps of 25 Watt, 50 Watt, 75 Watt,

until Respiratory Quotient RQ > 1.1

4)Cool down: 2 minutes bicycling without load

5)Recording of final haemodynamic values at rest after 5 minutes The patients were not informed if the control system in active or not and the sequence of the tests were changed at each patient.

Monitoring of the patients at catheter ergometry: 1.Swan-Ganz-Catheter: CVP; PA; PCWP; CO



2.Invasive Arterial Blood Pressure Measurement

3.Sensormedics®: Respiratory Quotient RQ

4.DeBakey VAD®: Flow; Speed; Current

5.Horizon®: ECG

The following diagram documents the benefit provided by the automatic control system during bicycle exercise. A severe relief of the pulmonary system could be observed in all four patient who underwent this test up to now. Further, the physical capacity could be increase in all patients.

For fast recovery of ventricular assist recipients and their sufficient adaptation to metabolic demand, appropriate monitoring and physiologically responsive control of the blood pumps are required. Within our clinical activities , our group has been deeply involved in the control of both membrane and rotary blood pumps. With membrane pumps, especially the specific situation in postoperative recovery, the balancing of two pumps for the bi-ventricular support and the use of venous oxygen content for pump adjustment has been investigated. With rotary pumps, the situation is even more complicated: Because of the lack of inherent adaptation to venous return and the difficulties to measure filling pressures with long-term stable means, the appropriate working point has to be detected by indirect means. We used conventional data analysis, fuzzy logic and neural networks to obtain indicators for optimal pump performance.