K9. Detection of Sleep Apnea Events using analysis of thoraco-abdominal excursion signals and adaptive neuro-fuzzy inference system (ANFIS)

Abstract

This paper describes an adaptive neuro-fuzzy inference system (ANFIS) for detection of Sleep Apnea Events using Thoracic and Abdominal Excursion signals. Mean amplitude sum analysis, and phase angle difference analysis using both Piecewise Linear Approximation (PLA) and phase difference measurements have been used to classify Normal, Obstructive Sleep Apnea (OSA), Hypopnea and Central Apnea events. A hybrid learning algorithm using a combination of Steepest Descent and Least Squares Estimation (LSE) was used to identify the parameters of ANFIS. The performance of ANFIS was evaluated in terms of training performance and classification accuracies and the results confirmed that the proposed ANFIS model has potential in classifying the Sleep Apnea events with an accuracy level of more than 95%.