Synchronization, Variability, and Nonlinearity Analysis: Applications to Physiological Time Series

In this thesis, we study synchronization, variability, nonlinearity analysis and their applications in physiological time series. For synchronization analysis, we explore both intensity and directionality of interaction. We propose a computational method to specify frequency ratio n : m, detect the moment when the frequency ratio changes, and also propose a novel method to measure the intensity of synchronization in n : m coupled system. Our method involves circular Poincare analysis of stroboscope and circular change point detection techniques. It is used to quantify interaction between respiration and ventilator recorded from lung injured rats. For directionality of coupling, we study Fourier series estimation approach and information-theoretic approach. We learn that Fourier series estimation approach is not practical because it is unknown how to automatically select an appropriate data segment for the estimation process. On the other hand, we show that the information theory approach simplifies the process of selecting a data segment to compute directionality of coupling. Therefore, it is appropriate for the physiological time series. The last topic in synchronization analysis is the study of neurodevelopment in neonates by evaluating synchronization in the EEG recorded from different regions of the brain. The result suggests that synchronization can be used to distinguish infants with different gestational ages. Specifically, based on intensity of synchronization in EEG, the late preterm group is more similar to the fullterm group than the mid preterm group. For the variability analysis, we propose a novel method to quantify heart rate variability (HRV) in ventilated rats after lung injury. We discover that periodicity of R-R interval is a key to distinguish difference of HRV in two ventilation groups. For nonlinearity analysis, we attempt to distinguish a nonlinear dynamical system from nonlinear transformation of a linear system. We investigate a nonlinear detection technique based on IAAFT surrogate and realize that its performance depends on both degree of nonlinearity of the transformation functions and nonlinear measures. Finally, we propose a novel technique to be used together with Barahona prediction method to fully distinguish various nonlinear schemes. Our method is based on residual analysis of a Volterra-Wiener-Korenberg model.

Name thesis:degree_name

Doctor of Philosophy

Level thesis:degree_level

doctoral

Discipline thesis:degree_discipline

EECS - System and Control Engineering

Grantor dc:publisher

Case Western Reserve University School of Graduate Studies

Year dc:date

2013

Author dc:creator

• Thungtong, Anurak

Contributors dc:contributor

• Loparo, Kenneth

• Electrical Engineering
• Biomedical Engineering
• Systems Design
• Statistics
• Synchronization
• Coupling direction
• Variability
• HRV
• Nonlinearity
• Physiological time series

Statement dc:rights

• unrestricted
• This thesis or dissertation is protected by copyright: all rights reserved. It may not be copied or redistributed beyond the terms of applicable copyright laws.

Language dc:language

English

Repository record dc:identifier

http://rave.ohiolink.edu/etdc/view?acc_num=case1364316597

OAI identifier oai:identifier

oai:etd.ohiolink.edu:case1364316597