Application of multiobjective neural predictive control to biventricular assistance using dual rotary blood pumps

Ng, Boon Chiang and Salamonsen, Robert F. and Gregory, Shaun D. and Stevens, Michael C. and Wu, Yi and Mansouri, Mahdi and Lovell, Nigel H. and Lim, Einly (2018) Application of multiobjective neural predictive control to biventricular assistance using dual rotary blood pumps. Biomedical Signal Processing and Control, 39. pp. 81-93. ISSN 1746-8094

Full text not available from this repository.
Official URL: https://doi.org/10.1016/j.bspc.2017.07.028

Abstract

Rotary blood pumps are used to provide mechanical circulatory support to the failing heart in patients who are ineligible or waiting for a transplant. One of the major challenges when implementing two rotary blood pumps for biventricular support is the difficulty in maintaining pulmonary and systemic circulatory volume balance. In this study, a novel multiobjective neural predictive controller (MONPC) hybridized with a preload-based Frank-Starling-like controller (PFS) has been proposed for a dual rotary blood pump biventricular assist device in two different configurations: PFS L -MONPC R and MONPC L -PFS R . The flow rate of one pump is regulated by PFS as a function of preload, while the other pump is controlled by MONPC, which is intended to meet cardiac demand, avoid pulmonary congestion and ventricular suction. A comparative assessment was performed between the proposed controllers and a Dual Independent Frank-Starling-like control system (DI-FS) as well as a constant speed controller. The numerical simulation results showed that MONPC L -PFS R helped unload the congested left ventricle while maintaining high cardiac output during exercise. In contrast, improper flow regulation by DI-FS has resulted in pulmonary congestion. During blood loss, PFS L -MONPC R delivered the lowest suction risk, as compared to the constant speed mode, which produced negative right ventricular preload. When sensor noise and time delays were introduced in the flow and end-diastolic pressure signals, the proposed controllers were able to respond with adequate robustness during the transition from rest to exercise. This study demonstrated that the proposed controllers are superior in matching the pump flow with the cardiac demand without causing hemodynamic instabilities.

Item Type: Article
Uncontrolled Keywords: Artificial neural network; Multiobjective optimization; Model predictive control; Biventricular assist device
Subjects: R Medicine
Divisions: Faculty of Engineering
Depositing User: Ms. Juhaida Abd Rahim
Date Deposited: 07 May 2019 07:57
Last Modified: 07 May 2019 07:57
URI: http://eprints.um.edu.my/id/eprint/21152

Actions (login required)

View Item View Item