Acoustic Breath Detection and Classification Modeling Respiratory Events

Abstract — A continuation of research into modeling airway events of patients undergoing sedation is described. Sounds recorded at the trachea were recorded and separated by means of a threshold algorithm. The threshold was determined by the expectation maximization algorithm on filtered data. A comparison between the respiratory rate of the threshold algorithm and that of the direct airflow measure is done. Classification of the audio airway events is discussed using both Neural Networks and Polynomial Classifiers. Future work will be discusse