Automated off-line cardiorespiratory event detection and validation
Ahmed Aoude · eScholarship@McGill (McGill) · 2006
Sleep apnea is a condition where breathing unexpectedly stops during sleep. This condition is a common medical problem affecting infants that can result in serious complications if left untreated. Anesthesia can increase episodes of post-operative sleep apnea in infants. Therefore, the monitoring of infants after surgery is of utmost importance. The standard for diagnosing apnea events remains the visual scoring of cardiorespiratory data by trained personnel. This process is time consuming and prone to human error. In this thesis, we present automated off-line algorithms for the detection of pauses, asynchrony and movement artifact in cardiorespiratory data. These algorithms were implemented in a new tool intended to replace the visual scoring process. The automated algorithms' effectiveness relative to visual scoring is presented. This comparison was achieved using a new visual scoring tool. Results presented in this thesis demonstrate that the developed methods are comparable to visual scoring, work with uncalibrated respiratory signals and provide quick, reliable and standardized analysis.