基于弹性管非线性模型盲辨识的中心动脉波无创动态重建研究

Central aortic pulse wave,which contains the information about cardiac dynamics and global circulation, is useful for diagnosis, treatment and prognosis of cardiovascualr diseases. If the central aortic pulse wave can be monitored in ambulatory, the variability of cardiovascular system can be detected more effectively at the earlier stage. Unfortunately, central aortic pulse wave only can be invasively measured. The analysis of the characteristics of cardiovascular system can not be dynamically achieved. It is reported that aortic pulse wave could be estimated by peripheral pulse waves. Existed estimation methods still assume the cardiovascular system is a time-invariant "black box" and the models can not be updated in real time. In this project, human vascular network is regarded as a single-input multiple-output (SIMO) system, employing multi-channel blind identification method to reconstruct the aortic pulse wave in real time. The input of the system is aortic pulse wave, and the outputs are multi-site peripheral pulse waves. Body area network is used to acquire multi-site peripheral pulse waves noninvasively. This project aims to solve the common scientific problem for blindly identifying the SIMO system with close or common zeros from the output signals with low SNR by fusing prior knowledge about the cardiovascular system and pulse wave propagation theory. Furthermore, the non-invasively monitoring of aortic pulse wave will provide some new theoretical basis and technical support for the disclosure of the characteristics of cardiovascular system when subject is with daily activities.

中心动脉波是心血管疾病的诊断、预防和治疗的重要依据。若在日常行为下对人体中心动脉波进行连续监测，就能更早、更有效地发现心血管系统的异常。但是，中心动脉波只能通过介入方式有创测得。目前，由体表脉搏波来估计中心动脉波的方法均假设系统为线性时不变的黑箱，没能将模型参数关联到血管机械特性和血流传播特征；而且，心血管系统本质是非线性的，所以通过线性模型不能有效地重建中心动脉波。本项目将日常行为下的人体心血管系统看成一单输入多输出的非线性时变弹性管系统，拟结合LS-SVM子空间法与时-频多尺度互相关法通过体表多点脉搏波来实时盲辨识各非线性弹性管通路的时变参数，并实时重建中心动脉波。最终，提供解决这类低信噪比、具有相近或公共零点的单输入多输出非线性时变系统的实时盲辨识共性问题的系统方法；还为分析血管时变机械特性及心血管调控能力提供量化评估，进而为揭示自然状态下心血管系统的调控机理提供理论依据和技术支持。

中心动脉波是心血管疾病的诊断、预防和治疗的重要依据。若在日常行为下对人体中心动脉波进行连续监测，就能更早、更有效地发现心血管系统的异常。但是，中心动脉波只能通过介入方式有创测得。目前，由体表脉搏波来估计中心动脉波的方法均假设系统为线性时不变系统，没有考虑其实变特性和个体差异；仅将心血管系统看做黑箱，未能将模型参数关联到血管机械特性和血流传播特征；心血管系统本质是非线性的，未能考虑其非线性成分，所以不能有效地重建中心动脉波。本项目将日常行为下的人体心血管系统看成一单输入多输出的非线性时变弹性管系统，结合子空间法与时-频多尺度互相关法通过体表多点脉搏波来在线盲辨识各非线性弹性管通路的时变参数，并实时重建中心动脉波。最终，提供解决这类低信噪比、具有相近或公共零点的单输入多输出非线性时变系统的实时盲辨识共性问题的系统方法；还为分析血管时变机械特性及心血管系统调控能力提供量化评估，进而为揭示自然状态下心血管系统的调控机理以及相关疾病的发生、发展提供理论依据与技术支持。