Methods for Detecting and Monitoring of Sleep Disordered Breathing in Children using Overnight Polysomnography

Abstract

Sleep is crucial for the health of every individual, especially children. One of the common causes of disturbed sleep in children is disordered breathing. Children who suffer from sleep disordered breathing are likely to have severe consequences for their physical growth, heart health and neuropsychological function. Sleep disordered breathing (SDB) comprises a spectrum of severity from a mild form of upper airway resistance syndrome (UARS) to severe form of obstructive sleep apnea syndrome (OSAS). While OSAS is considered clinically significant, UARS and its health consequences have been underestimated. The most common treatment for OSAS in children is adenotonsillectomy. However, breathing disturbances related to UARS may persist even after adenotonsillectomy. The current diagnostic marker for OSAS, the Apnea-Hypopnea Index (AHI) often overlooks the less severe conditions of breathing disturbances. Therefore, the research objective of this thesis is to investigate the new alternative markers for SDB in children using non-invasive physiological measurements, such as thoracoabdominal signals and the photoplethysmogram. As the body experiences an array of complex changes, specifically in respiratory and autonomic nervous system activation during breathing disturbances, advanced signal processing and analysis techniques were used to identify the physiological variables that could reflect changes in those systems in children with SDB. Thoraco-abdominal asynchrony (TAA), heart period (HP) and pulse wave amplitude (PWA) were the three physiological variables were investigated. A total of five studies were conducted on two high-quality clinical research datasets to test the potential of the proposed physiological variables to effectively identify children with SDB. In the thesis: 1) Hilbert transform was applied for TAA estimation on the childhood adenotonsillectomy trial (CHAT) dataset; 2) symbolic dynamic analysis on HP was used to assess the effect of adenotonsillectomy on autonomic activations in children with SDB; 3) the conventional method of estimating PWA was combined with joint symbolic analysis of PWA and HP to analyse the effect of SDB on autonomic activation compared to healthy controls; 4) to improve the performance of the previous PWA measurement technique, a more robust and simpler method was proposed to estimate PWA using a simple envelope method, and a more extensive dynamic analysis method was created to capture more complete information; and 5) adding TAA and HP information with AHI, unsupervised machine learning method K-means clustering and linear discriminant analysis were used to discover the pathophysiology nature difference of children with SDB in CHAT dataset. The main results from this thesis suggest that children with SDB have higher values in all three physiological variables, which indicates a high respiratory effort and elevated frequency of autonomic activation. Adenotonsillectomy showed to reverse the effects on these physiological variables, suggesting it assisted in the reduce of pathophysiological symptoms in those children. Interestingly, TAA was found inversely correlated with quality of life and unreported baseline difference in HP in children who had their AHI normalised spontaneously. These findings further indicate the limitation of AHI as the only marker for paediatric sleep disordered breathing. By combining the TAA and HP information with AHI, the alternative proposed diagnosing approach could help doctors predict who may benefit from adenotonsillectomy or not. In conclusion, this thesis provides new evidence that TAA, HP and PWA can provide additional information and may yield more effective markers for diagnosing paediatric sleep disordered breathing.