Data analysis in respiratory medicine : model selection and classification methods applied to clinical diagnostics

This thesis deals with the development of methods from the field of <br>data analysis for the advancement of clinical diagnostics in respiratory <br>medicine. <br> <br>Part I first exposes the clinical problem of incomplete forced expiration manoeuvres, an inconvenience that frequently arises when the lung function of patients with impaired respiration is investigated. We developed a model-based approach that provides estimates of vital capacity by analysing volume-time curves from uncompleted manoeuvres. The method was applied to a large number of clinical data sets and yielded unbiased estimates of vital capacity. <br> <br>Part II addresses the assessment of sleep-related breathing disorders by nocturnal monitoring of respiratory signals. Diagnosis of sleep-disordered breathing nowadays is carried out by polysomnography, i.e. by measurement of a multitude of neurologic and respiratory signals throughout the night in a sleep laboratory, and by time-consuming visual analysis of these data. The most important aspect of the analysis of polysomnographic data with respect to the differential diagnosis of sleep-related breathing disorders is the identification and classification of different kinds of respiratory events during sleep, mainly obstructive, mixed or central apnoeas and hypopnoeas. <br> <br>We addressed the potential of acoustic respiratory input impedance for diagnostic purposes. We present a method to detect and classify episodes of disturbed respiration during sleep by automatic analysis of two signals that are obtained simultaneously via a nasal mask: mask pressure and a signal closely related to acoustic input impedance of the respiratory system, measured by forced oscillation technique (FOT). The agreement between visual analysis of polysomnographic recordings and automatic analysis of mask pressure and FOT time series was overall as good as the agreement between independent clinical analyses of polysomnograms carried out by two experts in sleep medicine.

Author dc:creator

• Steltner, Holger

Contributors dc:contributor

• Honerkamp, Josef

• Modellselektion

Repository record source_url

https://freidok.uni-freiburg.de/data/324

OAI identifier oai:identifier

oai:freidok.uni-freiburg.de:324