Identification of Asthma Severity Using Wavelet Transform and K-nearest-neighbour Classifier

Abstract

This study aimed to classify wheeze sound according to asthma patient severity levels (mild, moderate and severe) using wavelet transform and K-nearest neighbour (KNN) classifier. Wheeze sounds were obtained from the lower lung base (LLB) and trachea of 55 asthmatic patients with different severity levels during tidal breathing manoeuvres. From the collected data, nine datasets were obtained based on the auscultation location, breath phases and / or their combination. From wheeze samples features has been extracted through 7th order wavelet transform. Furthermore, the K-nearest neighbour (KNN) classifier was implemented to classify severity levels. The highest positive predictive value (PPV) obtain for mild, moderate and severe samples is 84%, 77% and 75% respectively with the trachea related dataset. The results of this study indicate that through wavelet transform and KNN classifier the severity level of the asthma patients can be classified with tidal breathing. In the comparison of location, trachea is more specific and predictor for the severity level of asthma patients. Furthermore all datasets behaves differently for the classification of mild, moderate and severe asthma patients. With the comparison of phase inspiratory-related and expiratory-related datasets performs almost equally.