Characterization of cardiorespiratory coupling via a variability-based multi-method approach Application to postural orthostatic tachycardia syndrome

Abstract There are several mechanisms responsible for the dynamical link between heart period (HP) and respiration (R), usually referred to as cardiorespiratory coupling (CRC). Historically, diverse signal processing techniques have been employed to study CRC from the spontaneous fluctuations of HP and respiration (R). The proposed tools differ in terms of rationale and implementation, capturing diverse aspects of CRC. In this review, we classify the existing methods and stress differences with the aim of proposing a variability-based multi-method approach to CRC evaluation. Ten methodologies for CRC estimation, namely, power spectral decomposition, traditional and causal squared coherence,\;information transfer, cross-conditional entropy, mixed prediction, Shannon entropy of the latency between heartbeat and inspiratory/expiratory onset, conditional entropy of the phase dynamics, synchrogram-based analysis, pulse-respiration quotient, and joint symbolic dynamics, are considered. The ability of these techniques was exemplified over recordings acquired from patients suffering from postural orthostatic tachycardia syndrome (POTS) and healthy controls. Analyses were performed at rest in the supine position (REST) and during head-up tilt (HUT). Although most of the methods indicated that at REST, the CRC was lower in POTS patients and decreased more evidently during HUT in POTS, peculiar differences stressed the complementary value of the approaches. The multiple perspectives provided by the variability-based multi-method approach to CRC evaluation help the characterization of a pathological state and/or the quantification of the effect of a postural challenge. The present work stresses the need for the application of multiple methods to derive a more complete evaluation of the CRC in humans.摘要 心动周期（heart period）与呼吸（respiration）之间的动态关联存在多种调控机制，通常被称为心肺耦合（cardiorespiratory coupling，简称CRC）。长期以来，研究者已采用多种信号处理技术，基于心动周期与呼吸的自发波动对心肺耦合展开研究。所提出的各类工具在理论依据与实现方式上各有侧重，能够捕捉心肺耦合的多维度特征。在本综述中，我们对现有心肺耦合分析方法进行分类并梳理其差异，旨在提出一种基于变异性分析的多方法联合评估方案。 本文共纳入十种心肺耦合估计算法，分别为功率谱分解（power spectral decomposition）、传统与因果平方相干分析（traditional and causal squared coherence）、信息传递（information transfer）、交叉条件熵（cross-conditional entropy）、混合预测（mixed prediction）、心跳与吸气/呼气起始时刻延迟的香农熵（Shannon entropy of the latency between heartbeat and inspiratory/expiratory onset）、相位动力学条件熵（conditional entropy of the phase dynamics）、基于同步图的分析（synchrogram-based analysis）、呼吸脉搏商值（pulse-respiration quotient）以及联合符号动力学（joint symbolic dynamics）。 我们以采集自体位性直立性心动过速综合征（postural orthostatic tachycardia syndrome，简称POTS）患者与健康对照者的生理记录数据为例，验证了这些技术的分析性能。分析分别在仰卧静息状态（REST）以及头高位倾斜试验（head-up tilt，简称HUT）过程中开展。尽管多数方法均显示，静息状态下POTS患者的心肺耦合水平低于健康对照，且在头高位倾斜试验过程中POTS患者的心肺耦合下降更为显著，但各方法间的特有差异也凸显了不同分析路径的互补价值。 基于变异性分析的多方法联合评估方案所提供的多维度视角，有助于病理状态的特征刻画与体位挑战效应的量化分析。本研究强调，需采用多种分析方法以实现人类心肺耦合的全面评估。