Design and intelligent control of mock circulation system to reproduce patient-specific physiological indexes. (September 2022)

• Record Type: Journal Article
• Authors: Li, Te; Li, Heng; Cui, Wenbo; Xie, Nan; Li, Xu; Wang, Yongqing
• Is Part Of: Biomedical signal processing and control. Volume 78(2022)
• Publication Date: 2022-09
• Subjects: Mock circulatory system -- Deep reinforce-ment learning -- Intelligent control -- Personalized simula-tion -- Cardiovascular device evaluation; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2022.103987
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: Cardiovascular devices require a broader and more accurate in vitro environment. Personalized mock circulatory system is a valuable tool for hemodynamic studies. Deep reinforcement learning performs well for in vitro complex control. System rapidly and accurately reproduces patient-specific arterial characteristics. Abstract: Mock circulatory system (MCS) is a practicable solution to the cardiovascular device evaluation and pathological study prior to in vivo testing. In order to enhance the patient-specific imitative ability of the MCS in vitro tests, a personalized MCS that simulates aortic pressure and mean flow for a patient-specific condition is developed. The designed MCS can be fully controlled automatically, which is consisted of left ventricle, aortic compliance chamber, systemic vascular resistance and reservoir. A soft actor-critic based intelligent control method for personalized in vitro simulation is proposed, which is strongly robust to the MCS complexity and uncertainty. The designed MCS with the proposed control method can simulate aortic pressure and flow of a personalized patient with high accuracy and fast adjusting speed. The MCS control experiments are conducted with several groups of typical cardiovascular patients' physiological indexes. The results show that the maximum deviations of aortic systolic and diastolic pressures from the target values are less than 2 mmHg, and the maximum deviation of aortic flow is less than 0.2 L/min. Therefore, the developed MCS is potential to the personalized in vitro evaluation of cardiovascular devices such as ventricular assist devices and heart valves.